Upload folder using huggingface_hub
Browse files- README.md +14 -14
- all_results.json +17 -17
- eval_results.json +13 -13
- model-00001-of-00004.safetensors +1 -1
- model-00002-of-00004.safetensors +1 -1
- model-00003-of-00004.safetensors +1 -1
- model-00004-of-00004.safetensors +1 -1
- train_results.json +5 -5
- trainer_state.json +1052 -1052
- training_args.bin +1 -1
README.md
CHANGED
|
@@ -19,15 +19,15 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 19 |
|
| 20 |
This model is a fine-tuned version of [meta-llama/Meta-Llama-3.1-8B-Instruct](https://huggingface.co/meta-llama/Meta-Llama-3.1-8B-Instruct) on the simonycl/llama3.1-ultrafeedback-annotate-armorm dataset.
|
| 21 |
It achieves the following results on the evaluation set:
|
| 22 |
-
- Loss: 0.
|
| 23 |
-
- Rewards/chosen: -3.
|
| 24 |
-
- Rewards/rejected: -5.
|
| 25 |
-
- Rewards/accuracies: 0.
|
| 26 |
-
- Rewards/margins: 1.
|
| 27 |
-
- Logps/rejected: -
|
| 28 |
-
- Logps/chosen: -
|
| 29 |
-
- Logits/rejected: -
|
| 30 |
-
- Logits/chosen: -
|
| 31 |
|
| 32 |
## Model description
|
| 33 |
|
|
@@ -47,14 +47,14 @@ More information needed
|
|
| 47 |
|
| 48 |
The following hyperparameters were used during training:
|
| 49 |
- learning_rate: 5e-07
|
| 50 |
-
- train_batch_size:
|
| 51 |
-
- eval_batch_size:
|
| 52 |
- seed: 42
|
| 53 |
- distributed_type: multi-GPU
|
| 54 |
- num_devices: 4
|
| 55 |
-
- gradient_accumulation_steps:
|
| 56 |
- total_train_batch_size: 128
|
| 57 |
-
- total_eval_batch_size:
|
| 58 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 59 |
- lr_scheduler_type: cosine
|
| 60 |
- lr_scheduler_warmup_ratio: 0.1
|
|
@@ -64,7 +64,7 @@ The following hyperparameters were used during training:
|
|
| 64 |
|
| 65 |
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
| 66 |
|:-------------:|:------:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
| 67 |
-
| 0.
|
| 68 |
|
| 69 |
|
| 70 |
### Framework versions
|
|
|
|
| 19 |
|
| 20 |
This model is a fine-tuned version of [meta-llama/Meta-Llama-3.1-8B-Instruct](https://huggingface.co/meta-llama/Meta-Llama-3.1-8B-Instruct) on the simonycl/llama3.1-ultrafeedback-annotate-armorm dataset.
|
| 21 |
It achieves the following results on the evaluation set:
|
| 22 |
+
- Loss: 0.3837
|
| 23 |
+
- Rewards/chosen: -3.2511
|
| 24 |
+
- Rewards/rejected: -5.1202
|
| 25 |
+
- Rewards/accuracies: 0.8644
|
| 26 |
+
- Rewards/margins: 1.8691
|
| 27 |
+
- Logps/rejected: -797.6878
|
| 28 |
+
- Logps/chosen: -602.0981
|
| 29 |
+
- Logits/rejected: -1.3603
|
| 30 |
+
- Logits/chosen: -1.3921
|
| 31 |
|
| 32 |
## Model description
|
| 33 |
|
|
|
|
| 47 |
|
| 48 |
The following hyperparameters were used during training:
|
| 49 |
- learning_rate: 5e-07
|
| 50 |
+
- train_batch_size: 1
|
| 51 |
+
- eval_batch_size: 1
|
| 52 |
- seed: 42
|
| 53 |
- distributed_type: multi-GPU
|
| 54 |
- num_devices: 4
|
| 55 |
+
- gradient_accumulation_steps: 32
|
| 56 |
- total_train_batch_size: 128
|
| 57 |
+
- total_eval_batch_size: 4
|
| 58 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 59 |
- lr_scheduler_type: cosine
|
| 60 |
- lr_scheduler_warmup_ratio: 0.1
|
|
|
|
| 64 |
|
| 65 |
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
| 66 |
|:-------------:|:------:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
| 67 |
+
| 0.4269 | 0.8444 | 400 | 0.3837 | -3.2511 | -5.1202 | 0.8644 | 1.8691 | -797.6878 | -602.0981 | -1.3603 | -1.3921 |
|
| 68 |
|
| 69 |
|
| 70 |
### Framework versions
|
all_results.json
CHANGED
|
@@ -1,22 +1,22 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch": 0.
|
| 3 |
-
"eval_logits/chosen": -
|
| 4 |
-
"eval_logits/rejected": -
|
| 5 |
-
"eval_logps/chosen": -
|
| 6 |
-
"eval_logps/rejected": -
|
| 7 |
-
"eval_loss": 0.
|
| 8 |
-
"eval_rewards/accuracies": 0.
|
| 9 |
-
"eval_rewards/chosen": -3.
|
| 10 |
-
"eval_rewards/margins":
|
| 11 |
-
"eval_rewards/rejected": -5.
|
| 12 |
-
"eval_runtime": 316.
|
| 13 |
"eval_samples": 1976,
|
| 14 |
-
"eval_samples_per_second": 6.
|
| 15 |
-
"eval_steps_per_second":
|
| 16 |
"total_flos": 0.0,
|
| 17 |
-
"train_loss": 0.
|
| 18 |
-
"train_runtime":
|
| 19 |
"train_samples": 60634,
|
| 20 |
-
"train_samples_per_second":
|
| 21 |
-
"train_steps_per_second": 0.
|
| 22 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 0.9984827495217362,
|
| 3 |
+
"eval_logits/chosen": -1.4033699035644531,
|
| 4 |
+
"eval_logits/rejected": -1.3698593378067017,
|
| 5 |
+
"eval_logps/chosen": -627.2847900390625,
|
| 6 |
+
"eval_logps/rejected": -837.4208374023438,
|
| 7 |
+
"eval_loss": 0.3820858299732208,
|
| 8 |
+
"eval_rewards/accuracies": 0.8663967847824097,
|
| 9 |
+
"eval_rewards/chosen": -3.5029525756835938,
|
| 10 |
+
"eval_rewards/margins": 2.0145938396453857,
|
| 11 |
+
"eval_rewards/rejected": -5.517546653747559,
|
| 12 |
+
"eval_runtime": 316.7033,
|
| 13 |
"eval_samples": 1976,
|
| 14 |
+
"eval_samples_per_second": 6.239,
|
| 15 |
+
"eval_steps_per_second": 1.56,
|
| 16 |
"total_flos": 0.0,
|
| 17 |
+
"train_loss": 0.4645486564767285,
|
| 18 |
+
"train_runtime": 24049.7915,
|
| 19 |
"train_samples": 60634,
|
| 20 |
+
"train_samples_per_second": 2.521,
|
| 21 |
+
"train_steps_per_second": 0.02
|
| 22 |
}
|
eval_results.json
CHANGED
|
@@ -1,16 +1,16 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch": 0.
|
| 3 |
-
"eval_logits/chosen": -
|
| 4 |
-
"eval_logits/rejected": -
|
| 5 |
-
"eval_logps/chosen": -
|
| 6 |
-
"eval_logps/rejected": -
|
| 7 |
-
"eval_loss": 0.
|
| 8 |
-
"eval_rewards/accuracies": 0.
|
| 9 |
-
"eval_rewards/chosen": -3.
|
| 10 |
-
"eval_rewards/margins":
|
| 11 |
-
"eval_rewards/rejected": -5.
|
| 12 |
-
"eval_runtime": 316.
|
| 13 |
"eval_samples": 1976,
|
| 14 |
-
"eval_samples_per_second": 6.
|
| 15 |
-
"eval_steps_per_second":
|
| 16 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 0.9984827495217362,
|
| 3 |
+
"eval_logits/chosen": -1.4033699035644531,
|
| 4 |
+
"eval_logits/rejected": -1.3698593378067017,
|
| 5 |
+
"eval_logps/chosen": -627.2847900390625,
|
| 6 |
+
"eval_logps/rejected": -837.4208374023438,
|
| 7 |
+
"eval_loss": 0.3820858299732208,
|
| 8 |
+
"eval_rewards/accuracies": 0.8663967847824097,
|
| 9 |
+
"eval_rewards/chosen": -3.5029525756835938,
|
| 10 |
+
"eval_rewards/margins": 2.0145938396453857,
|
| 11 |
+
"eval_rewards/rejected": -5.517546653747559,
|
| 12 |
+
"eval_runtime": 316.7033,
|
| 13 |
"eval_samples": 1976,
|
| 14 |
+
"eval_samples_per_second": 6.239,
|
| 15 |
+
"eval_steps_per_second": 1.56
|
| 16 |
}
|
model-00001-of-00004.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4976698672
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c9ba9f8b42762f10643f8def0ec6cfddc799529412cca31ac9f19337c5249491
|
| 3 |
size 4976698672
|
model-00002-of-00004.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4999802720
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6acb42a4e8fa15ce25a4843b4ca4e825d92626a38e32dfaad5076e605dc736ec
|
| 3 |
size 4999802720
|
model-00003-of-00004.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4915916176
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6a8af3d5fba0a84018754fe35dc14fde7ed61b7dc4799f667dd630e0798ab82d
|
| 3 |
size 4915916176
|
model-00004-of-00004.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1168138808
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8a788bcf7589d73a13404a18ea27862b30c70bffb1514bb221b7aa5aa20773ae
|
| 3 |
size 1168138808
|
train_results.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch": 0.
|
| 3 |
"total_flos": 0.0,
|
| 4 |
-
"train_loss": 0.
|
| 5 |
-
"train_runtime":
|
| 6 |
"train_samples": 60634,
|
| 7 |
-
"train_samples_per_second":
|
| 8 |
-
"train_steps_per_second": 0.
|
| 9 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 0.9984827495217362,
|
| 3 |
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.4645486564767285,
|
| 5 |
+
"train_runtime": 24049.7915,
|
| 6 |
"train_samples": 60634,
|
| 7 |
+
"train_samples_per_second": 2.521,
|
| 8 |
+
"train_steps_per_second": 0.02
|
| 9 |
}
|
trainer_state.json
CHANGED
|
@@ -1,7 +1,7 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 0.
|
| 5 |
"eval_steps": 400,
|
| 6 |
"global_step": 473,
|
| 7 |
"is_hyper_param_search": false,
|
|
@@ -9,13 +9,13 @@
|
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
-
"epoch": 0.
|
| 13 |
-
"grad_norm":
|
| 14 |
"learning_rate": 1.0416666666666666e-08,
|
| 15 |
-
"logits/chosen": -
|
| 16 |
-
"logits/rejected": -
|
| 17 |
-
"logps/chosen": -
|
| 18 |
-
"logps/rejected": -
|
| 19 |
"loss": 0.6931,
|
| 20 |
"rewards/accuracies": 0.0,
|
| 21 |
"rewards/chosen": 0.0,
|
|
@@ -24,1439 +24,1439 @@
|
|
| 24 |
"step": 1
|
| 25 |
},
|
| 26 |
{
|
| 27 |
-
"epoch": 0.
|
| 28 |
-
"grad_norm": 4.
|
| 29 |
"learning_rate": 5.208333333333333e-08,
|
| 30 |
-
"logits/chosen": -
|
| 31 |
-
"logits/rejected": -
|
| 32 |
-
"logps/chosen": -
|
| 33 |
-
"logps/rejected": -
|
| 34 |
-
"loss": 0.
|
| 35 |
-
"rewards/accuracies": 0.
|
| 36 |
-
"rewards/chosen": 0.
|
| 37 |
-
"rewards/margins": 0.
|
| 38 |
-
"rewards/rejected":
|
| 39 |
"step": 5
|
| 40 |
},
|
| 41 |
{
|
| 42 |
-
"epoch": 0.
|
| 43 |
-
"grad_norm":
|
| 44 |
"learning_rate": 1.0416666666666667e-07,
|
| 45 |
-
"logits/chosen": -1.
|
| 46 |
-
"logits/rejected": -
|
| 47 |
-
"logps/chosen": -
|
| 48 |
-
"logps/rejected": -
|
| 49 |
"loss": 0.6933,
|
| 50 |
-
"rewards/accuracies": 0.
|
| 51 |
-
"rewards/chosen": 0.
|
| 52 |
-
"rewards/margins":
|
| 53 |
-
"rewards/rejected": 0.
|
| 54 |
"step": 10
|
| 55 |
},
|
| 56 |
{
|
| 57 |
-
"epoch": 0.
|
| 58 |
-
"grad_norm": 4.
|
| 59 |
"learning_rate": 1.5624999999999999e-07,
|
| 60 |
-
"logits/chosen": -
|
| 61 |
-
"logits/rejected": -
|
| 62 |
-
"logps/chosen": -
|
| 63 |
-
"logps/rejected": -
|
| 64 |
-
"loss": 0.
|
| 65 |
-
"rewards/accuracies": 0.
|
| 66 |
-
"rewards/chosen": 0.
|
| 67 |
-
"rewards/margins":
|
| 68 |
-
"rewards/rejected": 0.
|
| 69 |
"step": 15
|
| 70 |
},
|
| 71 |
{
|
| 72 |
-
"epoch": 0.
|
| 73 |
-
"grad_norm": 4.
|
| 74 |
"learning_rate": 2.0833333333333333e-07,
|
| 75 |
-
"logits/chosen": -
|
| 76 |
-
"logits/rejected": -
|
| 77 |
-
"logps/chosen": -
|
| 78 |
-
"logps/rejected": -
|
| 79 |
-
"loss": 0.
|
| 80 |
-
"rewards/accuracies": 0.
|
| 81 |
-
"rewards/chosen": -0.
|
| 82 |
-
"rewards/margins":
|
| 83 |
-
"rewards/rejected": 0.
|
| 84 |
"step": 20
|
| 85 |
},
|
| 86 |
{
|
| 87 |
-
"epoch": 0.
|
| 88 |
-
"grad_norm": 4.
|
| 89 |
"learning_rate": 2.604166666666667e-07,
|
| 90 |
-
"logits/chosen": -
|
| 91 |
-
"logits/rejected": -
|
| 92 |
-
"logps/chosen": -
|
| 93 |
-
"logps/rejected": -
|
| 94 |
-
"loss": 0.
|
| 95 |
-
"rewards/accuracies": 0.
|
| 96 |
-
"rewards/chosen": -0.
|
| 97 |
-
"rewards/margins": 0.
|
| 98 |
-
"rewards/rejected": -0.
|
| 99 |
"step": 25
|
| 100 |
},
|
| 101 |
{
|
| 102 |
-
"epoch": 0.
|
| 103 |
-
"grad_norm": 4.
|
| 104 |
"learning_rate": 3.1249999999999997e-07,
|
| 105 |
-
"logits/chosen": -
|
| 106 |
-
"logits/rejected": -
|
| 107 |
-
"logps/chosen": -
|
| 108 |
-
"logps/rejected": -
|
| 109 |
-
"loss": 0.
|
| 110 |
-
"rewards/accuracies": 0.
|
| 111 |
-
"rewards/chosen": -0.
|
| 112 |
-
"rewards/margins": 0.
|
| 113 |
-
"rewards/rejected": -0.
|
| 114 |
"step": 30
|
| 115 |
},
|
| 116 |
{
|
| 117 |
-
"epoch": 0.
|
| 118 |
-
"grad_norm": 4.
|
| 119 |
"learning_rate": 3.645833333333333e-07,
|
| 120 |
-
"logits/chosen": -
|
| 121 |
-
"logits/rejected": -
|
| 122 |
-
"logps/chosen": -
|
| 123 |
-
"logps/rejected": -
|
| 124 |
-
"loss": 0.
|
| 125 |
-
"rewards/accuracies": 0.
|
| 126 |
-
"rewards/chosen": -0.
|
| 127 |
-
"rewards/margins": 0.
|
| 128 |
-
"rewards/rejected": -0.
|
| 129 |
"step": 35
|
| 130 |
},
|
| 131 |
{
|
| 132 |
-
"epoch": 0.
|
| 133 |
-
"grad_norm": 4.
|
| 134 |
"learning_rate": 4.1666666666666667e-07,
|
| 135 |
-
"logits/chosen": -
|
| 136 |
-
"logits/rejected": -
|
| 137 |
-
"logps/chosen": -
|
| 138 |
-
"logps/rejected": -
|
| 139 |
-
"loss": 0.
|
| 140 |
-
"rewards/accuracies": 0.
|
| 141 |
-
"rewards/chosen": -0.
|
| 142 |
-
"rewards/margins": 0.
|
| 143 |
-
"rewards/rejected": -0.
|
| 144 |
"step": 40
|
| 145 |
},
|
| 146 |
{
|
| 147 |
-
"epoch": 0.
|
| 148 |
-
"grad_norm": 4.
|
| 149 |
"learning_rate": 4.6874999999999996e-07,
|
| 150 |
-
"logits/chosen": -
|
| 151 |
-
"logits/rejected": -
|
| 152 |
-
"logps/chosen": -
|
| 153 |
-
"logps/rejected": -
|
| 154 |
-
"loss": 0.
|
| 155 |
-
"rewards/accuracies": 0.
|
| 156 |
-
"rewards/chosen": -0.
|
| 157 |
-
"rewards/margins": 0.
|
| 158 |
-
"rewards/rejected": -0.
|
| 159 |
"step": 45
|
| 160 |
},
|
| 161 |
{
|
| 162 |
-
"epoch": 0.
|
| 163 |
-
"grad_norm": 4.
|
| 164 |
"learning_rate": 4.999726797933858e-07,
|
| 165 |
-
"logits/chosen": -1.
|
| 166 |
-
"logits/rejected": -
|
| 167 |
-
"logps/chosen": -
|
| 168 |
-
"logps/rejected": -
|
| 169 |
-
"loss": 0.
|
| 170 |
-
"rewards/accuracies": 0.
|
| 171 |
-
"rewards/chosen": -0.
|
| 172 |
-
"rewards/margins": 0.
|
| 173 |
-
"rewards/rejected": -0.
|
| 174 |
"step": 50
|
| 175 |
},
|
| 176 |
{
|
| 177 |
-
"epoch": 0.
|
| 178 |
-
"grad_norm": 6.
|
| 179 |
"learning_rate": 4.99665396039775e-07,
|
| 180 |
-
"logits/chosen": -1.
|
| 181 |
-
"logits/rejected": -
|
| 182 |
-
"logps/chosen": -
|
| 183 |
-
"logps/rejected": -
|
| 184 |
-
"loss": 0.
|
| 185 |
-
"rewards/accuracies": 0.
|
| 186 |
-
"rewards/chosen": -0.
|
| 187 |
-
"rewards/margins": 0.
|
| 188 |
-
"rewards/rejected": -0.
|
| 189 |
"step": 55
|
| 190 |
},
|
| 191 |
{
|
| 192 |
-
"epoch": 0.
|
| 193 |
-
"grad_norm": 7.
|
| 194 |
"learning_rate": 4.99017099386437e-07,
|
| 195 |
-
"logits/chosen": -
|
| 196 |
-
"logits/rejected": -1.
|
| 197 |
-
"logps/chosen": -
|
| 198 |
-
"logps/rejected": -
|
| 199 |
-
"loss": 0.
|
| 200 |
"rewards/accuracies": 0.75,
|
| 201 |
-
"rewards/chosen": -0.
|
| 202 |
-
"rewards/margins": 0.
|
| 203 |
-
"rewards/rejected": -0.
|
| 204 |
"step": 60
|
| 205 |
},
|
| 206 |
{
|
| 207 |
-
"epoch": 0.
|
| 208 |
-
"grad_norm":
|
| 209 |
"learning_rate": 4.980286753286194e-07,
|
| 210 |
-
"logits/chosen": -
|
| 211 |
-
"logits/rejected": -
|
| 212 |
-
"logps/chosen": -
|
| 213 |
-
"logps/rejected": -
|
| 214 |
-
"loss": 0.
|
| 215 |
-
"rewards/accuracies": 0.
|
| 216 |
-
"rewards/chosen": -1.
|
| 217 |
-
"rewards/margins": 0.
|
| 218 |
-
"rewards/rejected": -1.
|
| 219 |
"step": 65
|
| 220 |
},
|
| 221 |
{
|
| 222 |
-
"epoch": 0.
|
| 223 |
-
"grad_norm":
|
| 224 |
"learning_rate": 4.967014739346915e-07,
|
| 225 |
-
"logits/chosen": -
|
| 226 |
-
"logits/rejected": -
|
| 227 |
-
"logps/chosen": -
|
| 228 |
-
"logps/rejected": -
|
| 229 |
-
"loss": 0.
|
| 230 |
-
"rewards/accuracies": 0.
|
| 231 |
-
"rewards/chosen": -
|
| 232 |
-
"rewards/margins": 0.
|
| 233 |
-
"rewards/rejected": -1.
|
| 234 |
"step": 70
|
| 235 |
},
|
| 236 |
{
|
| 237 |
-
"epoch": 0.
|
| 238 |
-
"grad_norm":
|
| 239 |
"learning_rate": 4.950373080021136e-07,
|
| 240 |
-
"logits/chosen": -
|
| 241 |
-
"logits/rejected": -
|
| 242 |
-
"logps/chosen": -
|
| 243 |
-
"logps/rejected": -
|
| 244 |
-
"loss": 0.
|
| 245 |
-
"rewards/accuracies": 0.
|
| 246 |
-
"rewards/chosen": -0.
|
| 247 |
-
"rewards/margins": 0.
|
| 248 |
-
"rewards/rejected": -
|
| 249 |
"step": 75
|
| 250 |
},
|
| 251 |
{
|
| 252 |
-
"epoch": 0.
|
| 253 |
-
"grad_norm":
|
| 254 |
"learning_rate": 4.930384505813737e-07,
|
| 255 |
-
"logits/chosen": -
|
| 256 |
-
"logits/rejected": -
|
| 257 |
-
"logps/chosen": -
|
| 258 |
-
"logps/rejected": -
|
| 259 |
-
"loss": 0.
|
| 260 |
-
"rewards/accuracies": 0.
|
| 261 |
-
"rewards/chosen": -1.
|
| 262 |
-
"rewards/margins": 0.
|
| 263 |
-
"rewards/rejected": -
|
| 264 |
"step": 80
|
| 265 |
},
|
| 266 |
{
|
| 267 |
-
"epoch": 0.
|
| 268 |
-
"grad_norm":
|
| 269 |
"learning_rate": 4.907076318712738e-07,
|
| 270 |
-
"logits/chosen": -
|
| 271 |
-
"logits/rejected": -
|
| 272 |
-
"logps/chosen": -
|
| 273 |
-
"logps/rejected": -
|
| 274 |
-
"loss": 0.
|
| 275 |
-
"rewards/accuracies": 0.
|
| 276 |
-
"rewards/chosen": -1.
|
| 277 |
-
"rewards/margins": 0.
|
| 278 |
-
"rewards/rejected": -2.
|
| 279 |
"step": 85
|
| 280 |
},
|
| 281 |
{
|
| 282 |
-
"epoch": 0.
|
| 283 |
-
"grad_norm":
|
| 284 |
"learning_rate": 4.88048035489807e-07,
|
| 285 |
-
"logits/chosen": -
|
| 286 |
-
"logits/rejected": -
|
| 287 |
-
"logps/chosen": -
|
| 288 |
-
"logps/rejected": -
|
| 289 |
-
"loss": 0.
|
| 290 |
-
"rewards/accuracies": 0.
|
| 291 |
-
"rewards/chosen": -1.
|
| 292 |
-
"rewards/margins": 0.
|
| 293 |
-
"rewards/rejected": -1.
|
| 294 |
"step": 90
|
| 295 |
},
|
| 296 |
{
|
| 297 |
-
"epoch": 0.
|
| 298 |
-
"grad_norm":
|
| 299 |
"learning_rate": 4.85063294125718e-07,
|
| 300 |
-
"logits/chosen": -
|
| 301 |
-
"logits/rejected": -
|
| 302 |
-
"logps/chosen": -
|
| 303 |
-
"logps/rejected": -
|
| 304 |
-
"loss": 0.
|
| 305 |
-
"rewards/accuracies": 0.
|
| 306 |
-
"rewards/chosen": -1.
|
| 307 |
-
"rewards/margins": 0.
|
| 308 |
-
"rewards/rejected": -
|
| 309 |
"step": 95
|
| 310 |
},
|
| 311 |
{
|
| 312 |
-
"epoch": 0.
|
| 313 |
-
"grad_norm": 11.
|
| 314 |
"learning_rate": 4.817574845766874e-07,
|
| 315 |
-
"logits/chosen": -
|
| 316 |
-
"logits/rejected": -
|
| 317 |
-
"logps/chosen": -
|
| 318 |
-
"logps/rejected": -
|
| 319 |
-
"loss": 0.
|
| 320 |
"rewards/accuracies": 0.7749999761581421,
|
| 321 |
-
"rewards/chosen": -1.
|
| 322 |
-
"rewards/margins": 0.
|
| 323 |
-
"rewards/rejected": -2.
|
| 324 |
"step": 100
|
| 325 |
},
|
| 326 |
{
|
| 327 |
-
"epoch": 0.
|
| 328 |
-
"grad_norm":
|
| 329 |
"learning_rate": 4.781351221809166e-07,
|
| 330 |
-
"logits/chosen": -
|
| 331 |
-
"logits/rejected": -
|
| 332 |
-
"logps/chosen": -
|
| 333 |
-
"logps/rejected": -
|
| 334 |
-
"loss": 0.
|
| 335 |
-
"rewards/accuracies": 0.
|
| 336 |
-
"rewards/chosen": -
|
| 337 |
-
"rewards/margins": 0.
|
| 338 |
-
"rewards/rejected": -2.
|
| 339 |
"step": 105
|
| 340 |
},
|
| 341 |
{
|
| 342 |
-
"epoch": 0.
|
| 343 |
-
"grad_norm":
|
| 344 |
"learning_rate": 4.742011546497182e-07,
|
| 345 |
-
"logits/chosen": -
|
| 346 |
-
"logits/rejected": -
|
| 347 |
-
"logps/chosen": -
|
| 348 |
-
"logps/rejected": -
|
| 349 |
-
"loss": 0.
|
| 350 |
-
"rewards/accuracies": 0.
|
| 351 |
-
"rewards/chosen": -1.
|
| 352 |
-
"rewards/margins": 0.
|
| 353 |
-
"rewards/rejected": -2.
|
| 354 |
"step": 110
|
| 355 |
},
|
| 356 |
{
|
| 357 |
-
"epoch": 0.
|
| 358 |
-
"grad_norm": 16.
|
| 359 |
"learning_rate": 4.6996095530953875e-07,
|
| 360 |
-
"logits/chosen": -
|
| 361 |
-
"logits/rejected": -
|
| 362 |
-
"logps/chosen": -
|
| 363 |
-
"logps/rejected": -
|
| 364 |
-
"loss": 0.
|
| 365 |
-
"rewards/accuracies": 0.
|
| 366 |
-
"rewards/chosen": -2.
|
| 367 |
-
"rewards/margins": 1.
|
| 368 |
-
"rewards/rejected": -3.
|
| 369 |
"step": 115
|
| 370 |
},
|
| 371 |
{
|
| 372 |
-
"epoch": 0.
|
| 373 |
-
"grad_norm":
|
| 374 |
"learning_rate": 4.654203157626399e-07,
|
| 375 |
-
"logits/chosen": -
|
| 376 |
-
"logits/rejected": -
|
| 377 |
-
"logps/chosen": -
|
| 378 |
-
"logps/rejected": -
|
| 379 |
-
"loss": 0.
|
| 380 |
-
"rewards/accuracies": 0.
|
| 381 |
-
"rewards/chosen": -2.
|
| 382 |
-
"rewards/margins": 1.
|
| 383 |
-
"rewards/rejected": -3.
|
| 384 |
"step": 120
|
| 385 |
},
|
| 386 |
{
|
| 387 |
-
"epoch": 0.
|
| 388 |
-
"grad_norm":
|
| 389 |
"learning_rate": 4.605854379764673e-07,
|
| 390 |
-
"logits/chosen": -
|
| 391 |
-
"logits/rejected": -
|
| 392 |
-
"logps/chosen": -
|
| 393 |
-
"logps/rejected": -
|
| 394 |
-
"loss": 0.
|
| 395 |
-
"rewards/accuracies": 0.
|
| 396 |
-
"rewards/chosen": -
|
| 397 |
-
"rewards/margins":
|
| 398 |
-
"rewards/rejected": -
|
| 399 |
"step": 125
|
| 400 |
},
|
| 401 |
{
|
| 402 |
-
"epoch": 0.
|
| 403 |
-
"grad_norm":
|
| 404 |
"learning_rate": 4.5546292581250857e-07,
|
| 405 |
-
"logits/chosen": -
|
| 406 |
-
"logits/rejected": -
|
| 407 |
-
"logps/chosen": -
|
| 408 |
-
"logps/rejected": -
|
| 409 |
-
"loss": 0.
|
| 410 |
-
"rewards/accuracies": 0.
|
| 411 |
-
"rewards/chosen": -
|
| 412 |
-
"rewards/margins":
|
| 413 |
-
"rewards/rejected": -
|
| 414 |
"step": 130
|
| 415 |
},
|
| 416 |
{
|
| 417 |
-
"epoch": 0.
|
| 418 |
-
"grad_norm":
|
| 419 |
"learning_rate": 4.5005977600621275e-07,
|
| 420 |
-
"logits/chosen": -
|
| 421 |
-
"logits/rejected": -
|
| 422 |
-
"logps/chosen": -
|
| 423 |
-
"logps/rejected": -
|
| 424 |
-
"loss": 0.
|
| 425 |
-
"rewards/accuracies": 0.
|
| 426 |
-
"rewards/chosen": -2.
|
| 427 |
-
"rewards/margins": 1.
|
| 428 |
-
"rewards/rejected": -
|
| 429 |
"step": 135
|
| 430 |
},
|
| 431 |
{
|
| 432 |
-
"epoch": 0.
|
| 433 |
-
"grad_norm":
|
| 434 |
"learning_rate": 4.443833686102919e-07,
|
| 435 |
-
"logits/chosen": -
|
| 436 |
-
"logits/rejected": -
|
| 437 |
-
"logps/chosen": -
|
| 438 |
-
"logps/rejected": -
|
| 439 |
-
"loss": 0.
|
| 440 |
-
"rewards/accuracies": 0.
|
| 441 |
-
"rewards/chosen": -1.
|
| 442 |
-
"rewards/margins":
|
| 443 |
-
"rewards/rejected": -
|
| 444 |
"step": 140
|
| 445 |
},
|
| 446 |
{
|
| 447 |
-
"epoch": 0.
|
| 448 |
-
"grad_norm":
|
| 449 |
"learning_rate": 4.384414569144561e-07,
|
| 450 |
-
"logits/chosen": -
|
| 451 |
-
"logits/rejected": -
|
| 452 |
-
"logps/chosen": -
|
| 453 |
-
"logps/rejected": -
|
| 454 |
-
"loss": 0.
|
| 455 |
-
"rewards/accuracies": 0.
|
| 456 |
-
"rewards/chosen": -
|
| 457 |
-
"rewards/margins": 1.
|
| 458 |
-
"rewards/rejected": -
|
| 459 |
"step": 145
|
| 460 |
},
|
| 461 |
{
|
| 462 |
-
"epoch": 0.
|
| 463 |
-
"grad_norm":
|
| 464 |
"learning_rate": 4.3224215685535287e-07,
|
| 465 |
-
"logits/chosen": -
|
| 466 |
-
"logits/rejected": -1.
|
| 467 |
-
"logps/chosen": -
|
| 468 |
-
"logps/rejected": -
|
| 469 |
-
"loss": 0.
|
| 470 |
-
"rewards/accuracies": 0.
|
| 471 |
-
"rewards/chosen": -2.
|
| 472 |
-
"rewards/margins": 1.
|
| 473 |
-
"rewards/rejected": -3.
|
| 474 |
"step": 150
|
| 475 |
},
|
| 476 |
{
|
| 477 |
-
"epoch": 0.
|
| 478 |
-
"grad_norm":
|
| 479 |
"learning_rate": 4.2579393593117364e-07,
|
| 480 |
-
"logits/chosen": -
|
| 481 |
-
"logits/rejected": -1.
|
| 482 |
-
"logps/chosen": -
|
| 483 |
-
"logps/rejected": -
|
| 484 |
-
"loss": 0.
|
| 485 |
-
"rewards/accuracies": 0.
|
| 486 |
-
"rewards/chosen": -2.
|
| 487 |
-
"rewards/margins": 1.
|
| 488 |
-
"rewards/rejected": -
|
| 489 |
"step": 155
|
| 490 |
},
|
| 491 |
{
|
| 492 |
-
"epoch": 0.
|
| 493 |
-
"grad_norm":
|
| 494 |
"learning_rate": 4.191056016360699e-07,
|
| 495 |
-
"logits/chosen": -
|
| 496 |
-
"logits/rejected": -
|
| 497 |
-
"logps/chosen": -
|
| 498 |
-
"logps/rejected": -
|
| 499 |
-
"loss": 0.
|
| 500 |
-
"rewards/accuracies": 0.
|
| 501 |
-
"rewards/chosen": -
|
| 502 |
-
"rewards/margins": 1.
|
| 503 |
-
"rewards/rejected": -
|
| 504 |
"step": 160
|
| 505 |
},
|
| 506 |
{
|
| 507 |
-
"epoch": 0.
|
| 508 |
-
"grad_norm":
|
| 509 |
"learning_rate": 4.121862894301754e-07,
|
| 510 |
-
"logits/chosen": -
|
| 511 |
-
"logits/rejected": -1.
|
| 512 |
-
"logps/chosen": -
|
| 513 |
-
"logps/rejected": -
|
| 514 |
-
"loss": 0.
|
| 515 |
-
"rewards/accuracies": 0.
|
| 516 |
-
"rewards/chosen": -2.
|
| 517 |
-
"rewards/margins": 1.
|
| 518 |
-
"rewards/rejected": -
|
| 519 |
"step": 165
|
| 520 |
},
|
| 521 |
{
|
| 522 |
-
"epoch": 0.
|
| 523 |
-
"grad_norm":
|
| 524 |
"learning_rate": 4.050454502616667e-07,
|
| 525 |
-
"logits/chosen": -
|
| 526 |
-
"logits/rejected": -
|
| 527 |
-
"logps/chosen": -
|
| 528 |
-
"logps/rejected": -
|
| 529 |
-
"loss": 0.
|
| 530 |
-
"rewards/accuracies": 0.
|
| 531 |
-
"rewards/chosen": -2.
|
| 532 |
-
"rewards/margins": 1.
|
| 533 |
-
"rewards/rejected": -4.
|
| 534 |
"step": 170
|
| 535 |
},
|
| 536 |
{
|
| 537 |
-
"epoch": 0.
|
| 538 |
-
"grad_norm":
|
| 539 |
"learning_rate": 3.976928376579047e-07,
|
| 540 |
-
"logits/chosen": -
|
| 541 |
-
"logits/rejected": -
|
| 542 |
-
"logps/chosen": -
|
| 543 |
-
"logps/rejected": -
|
| 544 |
-
"loss": 0.
|
| 545 |
-
"rewards/accuracies": 0.
|
| 546 |
-
"rewards/chosen": -2.
|
| 547 |
-
"rewards/margins": 1.
|
| 548 |
-
"rewards/rejected": -
|
| 549 |
"step": 175
|
| 550 |
},
|
| 551 |
{
|
| 552 |
-
"epoch": 0.
|
| 553 |
-
"grad_norm": 15.
|
| 554 |
"learning_rate": 3.9013849440328945e-07,
|
| 555 |
-
"logits/chosen": -
|
| 556 |
-
"logits/rejected": -
|
| 557 |
-
"logps/chosen": -
|
| 558 |
-
"logps/rejected": -
|
| 559 |
-
"loss": 0.
|
| 560 |
-
"rewards/accuracies": 0.
|
| 561 |
-
"rewards/chosen": -2.
|
| 562 |
-
"rewards/margins": 1.
|
| 563 |
-
"rewards/rejected": -
|
| 564 |
"step": 180
|
| 565 |
},
|
| 566 |
{
|
| 567 |
-
"epoch": 0.
|
| 568 |
-
"grad_norm":
|
| 569 |
"learning_rate": 3.8239273882202473e-07,
|
| 570 |
-
"logits/chosen": -
|
| 571 |
-
"logits/rejected": -
|
| 572 |
-
"logps/chosen": -
|
| 573 |
-
"logps/rejected": -
|
| 574 |
-
"loss": 0.
|
| 575 |
-
"rewards/accuracies": 0.
|
| 576 |
-
"rewards/chosen": -2.
|
| 577 |
-
"rewards/margins": 1.
|
| 578 |
-
"rewards/rejected": -3.
|
| 579 |
"step": 185
|
| 580 |
},
|
| 581 |
{
|
| 582 |
-
"epoch": 0.
|
| 583 |
-
"grad_norm":
|
| 584 |
"learning_rate": 3.7446615068452804e-07,
|
| 585 |
-
"logits/chosen": -
|
| 586 |
-
"logits/rejected": -
|
| 587 |
-
"logps/chosen": -
|
| 588 |
-
"logps/rejected": -
|
| 589 |
-
"loss": 0.
|
| 590 |
"rewards/accuracies": 0.831250011920929,
|
| 591 |
-
"rewards/chosen": -2.
|
| 592 |
-
"rewards/margins": 1.
|
| 593 |
-
"rewards/rejected": -3.
|
| 594 |
"step": 190
|
| 595 |
},
|
| 596 |
{
|
| 597 |
-
"epoch": 0.
|
| 598 |
-
"grad_norm": 16.
|
| 599 |
"learning_rate": 3.6636955675673743e-07,
|
| 600 |
-
"logits/chosen": -
|
| 601 |
-
"logits/rejected": -
|
| 602 |
-
"logps/chosen": -
|
| 603 |
-
"logps/rejected": -
|
| 604 |
-
"loss": 0.
|
| 605 |
-
"rewards/accuracies": 0.
|
| 606 |
-
"rewards/chosen": -2.
|
| 607 |
-
"rewards/margins": 1.
|
| 608 |
-
"rewards/rejected": -
|
| 609 |
"step": 195
|
| 610 |
},
|
| 611 |
{
|
| 612 |
-
"epoch": 0.
|
| 613 |
-
"grad_norm":
|
| 614 |
"learning_rate": 3.5811401601205093e-07,
|
| 615 |
-
"logits/chosen": -
|
| 616 |
-
"logits/rejected": -
|
| 617 |
-
"logps/chosen": -
|
| 618 |
-
"logps/rejected": -
|
| 619 |
-
"loss": 0.
|
| 620 |
-
"rewards/accuracies": 0.
|
| 621 |
-
"rewards/chosen": -2.
|
| 622 |
-
"rewards/margins": 1.
|
| 623 |
-
"rewards/rejected": -4.
|
| 624 |
"step": 200
|
| 625 |
},
|
| 626 |
{
|
| 627 |
-
"epoch": 0.
|
| 628 |
-
"grad_norm":
|
| 629 |
"learning_rate": 3.497108045260995e-07,
|
| 630 |
-
"logits/chosen": -
|
| 631 |
-
"logits/rejected": -
|
| 632 |
-
"logps/chosen": -
|
| 633 |
-
"logps/rejected": -
|
| 634 |
-
"loss": 0.
|
| 635 |
-
"rewards/accuracies": 0.
|
| 636 |
-
"rewards/chosen": -2.
|
| 637 |
-
"rewards/margins": 1.
|
| 638 |
-
"rewards/rejected": -3.
|
| 639 |
"step": 205
|
| 640 |
},
|
| 641 |
{
|
| 642 |
-
"epoch": 0.
|
| 643 |
-
"grad_norm":
|
| 644 |
"learning_rate": 3.411714000749838e-07,
|
| 645 |
-
"logits/chosen": -
|
| 646 |
-
"logits/rejected": -
|
| 647 |
-
"logps/chosen": -
|
| 648 |
-
"logps/rejected": -
|
| 649 |
-
"loss": 0.
|
| 650 |
-
"rewards/accuracies": 0.
|
| 651 |
-
"rewards/chosen": -2.
|
| 652 |
-
"rewards/margins": 1.
|
| 653 |
-
"rewards/rejected": -4.
|
| 654 |
"step": 210
|
| 655 |
},
|
| 656 |
{
|
| 657 |
-
"epoch": 0.
|
| 658 |
-
"grad_norm":
|
| 659 |
"learning_rate": 3.3250746645801287e-07,
|
| 660 |
-
"logits/chosen": -
|
| 661 |
-
"logits/rejected": -
|
| 662 |
-
"logps/chosen": -
|
| 663 |
-
"logps/rejected": -
|
| 664 |
-
"loss": 0.
|
| 665 |
-
"rewards/accuracies": 0.
|
| 666 |
-
"rewards/chosen": -3.
|
| 667 |
-
"rewards/margins":
|
| 668 |
-
"rewards/rejected": -5.
|
| 669 |
"step": 215
|
| 670 |
},
|
| 671 |
{
|
| 672 |
-
"epoch": 0.
|
| 673 |
-
"grad_norm":
|
| 674 |
"learning_rate": 3.237308375663571e-07,
|
| 675 |
-
"logits/chosen": -
|
| 676 |
-
"logits/rejected": -
|
| 677 |
-
"logps/chosen": -
|
| 678 |
-
"logps/rejected": -
|
| 679 |
-
"loss": 0.
|
| 680 |
-
"rewards/accuracies": 0.
|
| 681 |
-
"rewards/chosen": -
|
| 682 |
-
"rewards/margins": 1.
|
| 683 |
-
"rewards/rejected": -
|
| 684 |
"step": 220
|
| 685 |
},
|
| 686 |
{
|
| 687 |
-
"epoch": 0.
|
| 688 |
-
"grad_norm":
|
| 689 |
"learning_rate": 3.148535012193767e-07,
|
| 690 |
-
"logits/chosen": -
|
| 691 |
-
"logits/rejected": -
|
| 692 |
-
"logps/chosen": -
|
| 693 |
-
"logps/rejected": -
|
| 694 |
-
"loss": 0.
|
| 695 |
-
"rewards/accuracies": 0.
|
| 696 |
-
"rewards/chosen": -
|
| 697 |
-
"rewards/margins":
|
| 698 |
-
"rewards/rejected": -
|
| 699 |
"step": 225
|
| 700 |
},
|
| 701 |
{
|
| 702 |
-
"epoch": 0.
|
| 703 |
-
"grad_norm":
|
| 704 |
"learning_rate": 3.0588758279070183e-07,
|
| 705 |
-
"logits/chosen": -
|
| 706 |
-
"logits/rejected": -
|
| 707 |
-
"logps/chosen": -
|
| 708 |
-
"logps/rejected": -
|
| 709 |
-
"loss": 0.
|
| 710 |
"rewards/accuracies": 0.824999988079071,
|
| 711 |
-
"rewards/chosen": -
|
| 712 |
-
"rewards/margins": 1.
|
| 713 |
-
"rewards/rejected": -
|
| 714 |
"step": 230
|
| 715 |
},
|
| 716 |
{
|
| 717 |
-
"epoch": 0.
|
| 718 |
-
"grad_norm": 16.
|
| 719 |
"learning_rate": 2.968453286464312e-07,
|
| 720 |
-
"logits/chosen": -
|
| 721 |
-
"logits/rejected": -
|
| 722 |
-
"logps/chosen": -
|
| 723 |
-
"logps/rejected": -
|
| 724 |
-
"loss": 0.
|
| 725 |
-
"rewards/accuracies": 0.
|
| 726 |
-
"rewards/chosen": -
|
| 727 |
-
"rewards/margins": 1.
|
| 728 |
-
"rewards/rejected": -
|
| 729 |
"step": 235
|
| 730 |
},
|
| 731 |
{
|
| 732 |
-
"epoch": 0.
|
| 733 |
-
"grad_norm":
|
| 734 |
"learning_rate": 2.8773908941806877e-07,
|
| 735 |
-
"logits/chosen": -
|
| 736 |
-
"logits/rejected": -
|
| 737 |
-
"logps/chosen": -
|
| 738 |
-
"logps/rejected": -
|
| 739 |
-
"loss": 0.
|
| 740 |
-
"rewards/accuracies": 0.
|
| 741 |
-
"rewards/chosen": -3.
|
| 742 |
-
"rewards/margins":
|
| 743 |
-
"rewards/rejected": -5.
|
| 744 |
"step": 240
|
| 745 |
},
|
| 746 |
{
|
| 747 |
-
"epoch": 0.
|
| 748 |
-
"grad_norm":
|
| 749 |
"learning_rate": 2.785813031330473e-07,
|
| 750 |
-
"logits/chosen": -
|
| 751 |
-
"logits/rejected": -
|
| 752 |
-
"logps/chosen": -
|
| 753 |
-
"logps/rejected": -
|
| 754 |
-
"loss": 0.
|
| 755 |
-
"rewards/accuracies": 0.
|
| 756 |
-
"rewards/chosen": -
|
| 757 |
-
"rewards/margins":
|
| 758 |
-
"rewards/rejected": -5.
|
| 759 |
"step": 245
|
| 760 |
},
|
| 761 |
{
|
| 762 |
-
"epoch": 0.
|
| 763 |
-
"grad_norm":
|
| 764 |
"learning_rate": 2.693844782258779e-07,
|
| 765 |
-
"logits/chosen": -
|
| 766 |
-
"logits/rejected": -1.
|
| 767 |
-
"logps/chosen": -
|
| 768 |
-
"logps/rejected": -
|
| 769 |
-
"loss": 0.
|
| 770 |
-
"rewards/accuracies": 0.
|
| 771 |
-
"rewards/chosen": -
|
| 772 |
-
"rewards/margins": 1.
|
| 773 |
-
"rewards/rejected": -
|
| 774 |
"step": 250
|
| 775 |
},
|
| 776 |
{
|
| 777 |
-
"epoch": 0.
|
| 778 |
-
"grad_norm":
|
| 779 |
"learning_rate": 2.601611764531342e-07,
|
| 780 |
-
"logits/chosen": -
|
| 781 |
-
"logits/rejected": -
|
| 782 |
-
"logps/chosen": -
|
| 783 |
-
"logps/rejected": -
|
| 784 |
-
"loss": 0.
|
| 785 |
-
"rewards/accuracies": 0.
|
| 786 |
-
"rewards/chosen": -
|
| 787 |
-
"rewards/margins":
|
| 788 |
-
"rewards/rejected": -
|
| 789 |
"step": 255
|
| 790 |
},
|
| 791 |
{
|
| 792 |
-
"epoch": 0.
|
| 793 |
-
"grad_norm":
|
| 794 |
"learning_rate": 2.5092399573560323e-07,
|
| 795 |
-
"logits/chosen": -
|
| 796 |
-
"logits/rejected": -
|
| 797 |
-
"logps/chosen": -
|
| 798 |
-
"logps/rejected": -
|
| 799 |
-
"loss": 0.
|
| 800 |
-
"rewards/accuracies": 0.
|
| 801 |
-
"rewards/chosen": -
|
| 802 |
-
"rewards/margins": 2.
|
| 803 |
-
"rewards/rejected": -
|
| 804 |
"step": 260
|
| 805 |
},
|
| 806 |
{
|
| 807 |
-
"epoch": 0.
|
| 808 |
-
"grad_norm":
|
| 809 |
"learning_rate": 2.4168555295104124e-07,
|
| 810 |
-
"logits/chosen": -
|
| 811 |
-
"logits/rejected": -
|
| 812 |
-
"logps/chosen": -
|
| 813 |
-
"logps/rejected": -
|
| 814 |
-
"loss": 0.
|
| 815 |
-
"rewards/accuracies": 0.
|
| 816 |
-
"rewards/chosen": -3.
|
| 817 |
-
"rewards/margins": 1.
|
| 818 |
-
"rewards/rejected": -5.
|
| 819 |
"step": 265
|
| 820 |
},
|
| 821 |
{
|
| 822 |
-
"epoch": 0.
|
| 823 |
-
"grad_norm":
|
| 824 |
"learning_rate": 2.3245846670103626e-07,
|
| 825 |
-
"logits/chosen": -
|
| 826 |
-
"logits/rejected": -
|
| 827 |
-
"logps/chosen": -
|
| 828 |
-
"logps/rejected": -
|
| 829 |
-
"loss": 0.
|
| 830 |
-
"rewards/accuracies": 0.
|
| 831 |
-
"rewards/chosen": -3.
|
| 832 |
-
"rewards/margins": 1.
|
| 833 |
-
"rewards/rejected": -
|
| 834 |
"step": 270
|
| 835 |
},
|
| 836 |
{
|
| 837 |
-
"epoch": 0.
|
| 838 |
-
"grad_norm":
|
| 839 |
"learning_rate": 2.232553400755159e-07,
|
| 840 |
-
"logits/chosen": -
|
| 841 |
-
"logits/rejected": -
|
| 842 |
-
"logps/chosen": -
|
| 843 |
-
"logps/rejected": -
|
| 844 |
-
"loss": 0.
|
| 845 |
-
"rewards/accuracies": 0.
|
| 846 |
-
"rewards/chosen": -3.
|
| 847 |
-
"rewards/margins": 2.
|
| 848 |
-
"rewards/rejected": -
|
| 849 |
"step": 275
|
| 850 |
},
|
| 851 |
{
|
| 852 |
-
"epoch": 0.
|
| 853 |
-
"grad_norm":
|
| 854 |
"learning_rate": 2.1408874343844294e-07,
|
| 855 |
-
"logits/chosen": -
|
| 856 |
-
"logits/rejected": -
|
| 857 |
-
"logps/chosen": -
|
| 858 |
-
"logps/rejected": -
|
| 859 |
-
"loss": 0.
|
| 860 |
-
"rewards/accuracies": 0.
|
| 861 |
-
"rewards/chosen": -
|
| 862 |
-
"rewards/margins": 2.
|
| 863 |
-
"rewards/rejected": -6.
|
| 864 |
"step": 280
|
| 865 |
},
|
| 866 |
{
|
| 867 |
-
"epoch": 0.
|
| 868 |
-
"grad_norm":
|
| 869 |
"learning_rate": 2.049711972582101e-07,
|
| 870 |
-
"logits/chosen": -
|
| 871 |
-
"logits/rejected": -
|
| 872 |
-
"logps/chosen": -
|
| 873 |
-
"logps/rejected": -
|
| 874 |
-
"loss": 0.
|
| 875 |
-
"rewards/accuracies": 0.
|
| 876 |
-
"rewards/chosen": -
|
| 877 |
-
"rewards/margins": 2.
|
| 878 |
-
"rewards/rejected": -6.
|
| 879 |
"step": 285
|
| 880 |
},
|
| 881 |
{
|
| 882 |
-
"epoch": 0.
|
| 883 |
-
"grad_norm":
|
| 884 |
"learning_rate": 1.9591515500618588e-07,
|
| 885 |
-
"logits/chosen": -
|
| 886 |
-
"logits/rejected": -
|
| 887 |
-
"logps/chosen": -
|
| 888 |
-
"logps/rejected": -
|
| 889 |
-
"loss": 0.
|
| 890 |
-
"rewards/accuracies": 0.
|
| 891 |
-
"rewards/chosen": -
|
| 892 |
-
"rewards/margins": 1.
|
| 893 |
-
"rewards/rejected": -5.
|
| 894 |
"step": 290
|
| 895 |
},
|
| 896 |
{
|
| 897 |
-
"epoch": 0.
|
| 898 |
-
"grad_norm":
|
| 899 |
"learning_rate": 1.8693298614677112e-07,
|
| 900 |
-
"logits/chosen": -
|
| 901 |
-
"logits/rejected": -
|
| 902 |
-
"logps/chosen": -
|
| 903 |
-
"logps/rejected": -
|
| 904 |
-
"loss": 0.
|
| 905 |
"rewards/accuracies": 0.875,
|
| 906 |
-
"rewards/chosen": -
|
| 907 |
-
"rewards/margins":
|
| 908 |
-
"rewards/rejected": -
|
| 909 |
"step": 295
|
| 910 |
},
|
| 911 |
{
|
| 912 |
-
"epoch": 0.
|
| 913 |
-
"grad_norm":
|
| 914 |
"learning_rate": 1.7803695924219814e-07,
|
| 915 |
-
"logits/chosen": -
|
| 916 |
-
"logits/rejected": -
|
| 917 |
-
"logps/chosen": -
|
| 918 |
-
"logps/rejected": -
|
| 919 |
-
"loss": 0.
|
| 920 |
-
"rewards/accuracies": 0.
|
| 921 |
-
"rewards/chosen": -3.
|
| 922 |
-
"rewards/margins":
|
| 923 |
-
"rewards/rejected": -5.
|
| 924 |
"step": 300
|
| 925 |
},
|
| 926 |
{
|
| 927 |
-
"epoch": 0.
|
| 928 |
-
"grad_norm":
|
| 929 |
"learning_rate": 1.6923922519515067e-07,
|
| 930 |
-
"logits/chosen": -
|
| 931 |
-
"logits/rejected": -
|
| 932 |
-
"logps/chosen": -
|
| 933 |
-
"logps/rejected": -
|
| 934 |
-
"loss": 0.
|
| 935 |
-
"rewards/accuracies": 0.
|
| 936 |
-
"rewards/chosen": -
|
| 937 |
-
"rewards/margins": 1.
|
| 938 |
-
"rewards/rejected": -4.
|
| 939 |
"step": 305
|
| 940 |
},
|
| 941 |
{
|
| 942 |
-
"epoch": 0.
|
| 943 |
-
"grad_norm":
|
| 944 |
"learning_rate": 1.605518006520924e-07,
|
| 945 |
-
"logits/chosen": -
|
| 946 |
-
"logits/rejected": -
|
| 947 |
-
"logps/chosen": -
|
| 948 |
-
"logps/rejected": -
|
| 949 |
-
"loss": 0.
|
| 950 |
-
"rewards/accuracies": 0.
|
| 951 |
-
"rewards/chosen": -3.
|
| 952 |
-
"rewards/margins": 2.
|
| 953 |
-
"rewards/rejected": -5.
|
| 954 |
"step": 310
|
| 955 |
},
|
| 956 |
{
|
| 957 |
-
"epoch": 0.
|
| 958 |
-
"grad_norm":
|
| 959 |
"learning_rate": 1.519865515899731e-07,
|
| 960 |
-
"logits/chosen": -
|
| 961 |
-
"logits/rejected": -
|
| 962 |
-
"logps/chosen": -
|
| 963 |
-
"logps/rejected": -
|
| 964 |
-
"loss": 0.
|
| 965 |
-
"rewards/accuracies": 0.
|
| 966 |
-
"rewards/chosen": -3.
|
| 967 |
-
"rewards/margins":
|
| 968 |
-
"rewards/rejected": -5.
|
| 969 |
"step": 315
|
| 970 |
},
|
| 971 |
{
|
| 972 |
-
"epoch": 0.
|
| 973 |
-
"grad_norm":
|
| 974 |
"learning_rate": 1.4355517710873182e-07,
|
| 975 |
-
"logits/chosen": -
|
| 976 |
-
"logits/rejected": -
|
| 977 |
-
"logps/chosen": -
|
| 978 |
-
"logps/rejected": -
|
| 979 |
-
"loss": 0.
|
| 980 |
-
"rewards/accuracies": 0.
|
| 981 |
-
"rewards/chosen": -3.
|
| 982 |
-
"rewards/margins": 2.
|
| 983 |
-
"rewards/rejected": -
|
| 984 |
"step": 320
|
| 985 |
},
|
| 986 |
{
|
| 987 |
-
"epoch": 0.
|
| 988 |
-
"grad_norm":
|
| 989 |
"learning_rate": 1.3526919345173318e-07,
|
| 990 |
-
"logits/chosen": -
|
| 991 |
-
"logits/rejected": -
|
| 992 |
-
"logps/chosen": -
|
| 993 |
-
"logps/rejected": -
|
| 994 |
-
"loss": 0.
|
| 995 |
"rewards/accuracies": 0.84375,
|
| 996 |
-
"rewards/chosen": -3.
|
| 997 |
-
"rewards/margins": 2.
|
| 998 |
-
"rewards/rejected": -5.
|
| 999 |
"step": 325
|
| 1000 |
},
|
| 1001 |
{
|
| 1002 |
-
"epoch": 0.
|
| 1003 |
-
"grad_norm":
|
| 1004 |
"learning_rate": 1.2713991827596443e-07,
|
| 1005 |
-
"logits/chosen": -
|
| 1006 |
-
"logits/rejected": -
|
| 1007 |
-
"logps/chosen": -
|
| 1008 |
-
"logps/rejected": -
|
| 1009 |
-
"loss": 0.
|
| 1010 |
-
"rewards/accuracies": 0.
|
| 1011 |
-
"rewards/chosen": -3.
|
| 1012 |
-
"rewards/margins": 1.
|
| 1013 |
-
"rewards/rejected": -5.
|
| 1014 |
"step": 330
|
| 1015 |
},
|
| 1016 |
{
|
| 1017 |
-
"epoch": 0.
|
| 1018 |
-
"grad_norm":
|
| 1019 |
"learning_rate": 1.191784551934773e-07,
|
| 1020 |
-
"logits/chosen": -
|
| 1021 |
-
"logits/rejected": -
|
| 1022 |
-
"logps/chosen": -
|
| 1023 |
-
"logps/rejected": -
|
| 1024 |
-
"loss": 0.
|
| 1025 |
-
"rewards/accuracies": 0.
|
| 1026 |
-
"rewards/chosen": -
|
| 1027 |
-
"rewards/margins":
|
| 1028 |
-
"rewards/rejected": -
|
| 1029 |
"step": 335
|
| 1030 |
},
|
| 1031 |
{
|
| 1032 |
-
"epoch": 0.
|
| 1033 |
-
"grad_norm":
|
| 1034 |
"learning_rate": 1.1139567860518953e-07,
|
| 1035 |
-
"logits/chosen": -
|
| 1036 |
-
"logits/rejected": -1.
|
| 1037 |
-
"logps/chosen": -
|
| 1038 |
-
"logps/rejected": -
|
| 1039 |
-
"loss": 0.
|
| 1040 |
-
"rewards/accuracies": 0.
|
| 1041 |
-
"rewards/chosen": -
|
| 1042 |
-
"rewards/margins": 1.
|
| 1043 |
-
"rewards/rejected": -
|
| 1044 |
"step": 340
|
| 1045 |
},
|
| 1046 |
{
|
| 1047 |
-
"epoch": 0.
|
| 1048 |
-
"grad_norm":
|
| 1049 |
"learning_rate": 1.0380221884776128e-07,
|
| 1050 |
-
"logits/chosen": -
|
| 1051 |
-
"logits/rejected": -
|
| 1052 |
-
"logps/chosen": -
|
| 1053 |
-
"logps/rejected": -
|
| 1054 |
-
"loss": 0.
|
| 1055 |
-
"rewards/accuracies": 0.
|
| 1056 |
-
"rewards/chosen": -2.
|
| 1057 |
-
"rewards/margins": 1.
|
| 1058 |
-
"rewards/rejected": -4.
|
| 1059 |
"step": 345
|
| 1060 |
},
|
| 1061 |
{
|
| 1062 |
-
"epoch": 0.
|
| 1063 |
-
"grad_norm":
|
| 1064 |
"learning_rate": 9.640844767383405e-08,
|
| 1065 |
-
"logits/chosen": -
|
| 1066 |
-
"logits/rejected": -
|
| 1067 |
-
"logps/chosen": -
|
| 1068 |
-
"logps/rejected": -
|
| 1069 |
-
"loss": 0.
|
| 1070 |
-
"rewards/accuracies": 0.
|
| 1071 |
-
"rewards/chosen": -2.
|
| 1072 |
-
"rewards/margins": 1.
|
| 1073 |
-
"rewards/rejected": -4.
|
| 1074 |
"step": 350
|
| 1075 |
},
|
| 1076 |
{
|
| 1077 |
-
"epoch": 0.
|
| 1078 |
-
"grad_norm": 22.
|
| 1079 |
"learning_rate": 8.922446408546378e-08,
|
| 1080 |
-
"logits/chosen": -
|
| 1081 |
-
"logits/rejected": -
|
| 1082 |
-
"logps/chosen": -
|
| 1083 |
-
"logps/rejected": -
|
| 1084 |
-
"loss": 0.
|
| 1085 |
-
"rewards/accuracies": 0.
|
| 1086 |
-
"rewards/chosen": -
|
| 1087 |
-
"rewards/margins":
|
| 1088 |
-
"rewards/rejected": -5.
|
| 1089 |
"step": 355
|
| 1090 |
},
|
| 1091 |
{
|
| 1092 |
-
"epoch": 0.
|
| 1093 |
-
"grad_norm":
|
| 1094 |
"learning_rate": 8.22600805400994e-08,
|
| 1095 |
-
"logits/chosen": -
|
| 1096 |
-
"logits/rejected": -
|
| 1097 |
-
"logps/chosen": -
|
| 1098 |
-
"logps/rejected": -
|
| 1099 |
-
"loss": 0.
|
| 1100 |
-
"rewards/accuracies": 0.
|
| 1101 |
-
"rewards/chosen": -
|
| 1102 |
-
"rewards/margins":
|
| 1103 |
-
"rewards/rejected": -
|
| 1104 |
"step": 360
|
| 1105 |
},
|
| 1106 |
{
|
| 1107 |
-
"epoch": 0.
|
| 1108 |
-
"grad_norm":
|
| 1109 |
"learning_rate": 7.552480954794558e-08,
|
| 1110 |
-
"logits/chosen": -
|
| 1111 |
-
"logits/rejected": -
|
| 1112 |
-
"logps/chosen": -
|
| 1113 |
-
"logps/rejected": -
|
| 1114 |
-
"loss": 0.
|
| 1115 |
-
"rewards/accuracies": 0.
|
| 1116 |
-
"rewards/chosen": -
|
| 1117 |
-
"rewards/margins": 1.
|
| 1118 |
-
"rewards/rejected": -
|
| 1119 |
"step": 365
|
| 1120 |
},
|
| 1121 |
{
|
| 1122 |
-
"epoch": 0.
|
| 1123 |
-
"grad_norm":
|
| 1124 |
"learning_rate": 6.902785067901854e-08,
|
| 1125 |
-
"logits/chosen": -
|
| 1126 |
-
"logits/rejected": -
|
| 1127 |
-
"logps/chosen": -
|
| 1128 |
-
"logps/rejected": -
|
| 1129 |
-
"loss": 0.
|
| 1130 |
-
"rewards/accuracies": 0.
|
| 1131 |
-
"rewards/chosen": -3.
|
| 1132 |
-
"rewards/margins":
|
| 1133 |
-
"rewards/rejected": -
|
| 1134 |
"step": 370
|
| 1135 |
},
|
| 1136 |
{
|
| 1137 |
-
"epoch": 0.
|
| 1138 |
-
"grad_norm":
|
| 1139 |
"learning_rate": 6.277807799763973e-08,
|
| 1140 |
-
"logits/chosen": -
|
| 1141 |
-
"logits/rejected": -
|
| 1142 |
-
"logps/chosen": -
|
| 1143 |
-
"logps/rejected": -
|
| 1144 |
-
"loss": 0.
|
| 1145 |
-
"rewards/accuracies": 0.
|
| 1146 |
-
"rewards/chosen": -3.
|
| 1147 |
-
"rewards/margins":
|
| 1148 |
-
"rewards/rejected": -5.
|
| 1149 |
"step": 375
|
| 1150 |
},
|
| 1151 |
{
|
| 1152 |
-
"epoch": 0.
|
| 1153 |
-
"grad_norm":
|
| 1154 |
"learning_rate": 5.678402794153145e-08,
|
| 1155 |
-
"logits/chosen": -
|
| 1156 |
-
"logits/rejected": -
|
| 1157 |
-
"logps/chosen": -
|
| 1158 |
-
"logps/rejected": -
|
| 1159 |
-
"loss": 0.
|
| 1160 |
-
"rewards/accuracies": 0.
|
| 1161 |
-
"rewards/chosen": -3.
|
| 1162 |
-
"rewards/margins":
|
| 1163 |
-
"rewards/rejected": -5.
|
| 1164 |
"step": 380
|
| 1165 |
},
|
| 1166 |
{
|
| 1167 |
-
"epoch": 0.
|
| 1168 |
-
"grad_norm":
|
| 1169 |
"learning_rate": 5.105388766206969e-08,
|
| 1170 |
-
"logits/chosen": -
|
| 1171 |
-
"logits/rejected": -
|
| 1172 |
-
"logps/chosen": -
|
| 1173 |
-
"logps/rejected": -
|
| 1174 |
-
"loss": 0.
|
| 1175 |
-
"rewards/accuracies": 0.
|
| 1176 |
-
"rewards/chosen": -
|
| 1177 |
-
"rewards/margins": 2.
|
| 1178 |
-
"rewards/rejected": -6.
|
| 1179 |
"step": 385
|
| 1180 |
},
|
| 1181 |
{
|
| 1182 |
-
"epoch": 0.
|
| 1183 |
-
"grad_norm":
|
| 1184 |
"learning_rate": 4.5595483841620484e-08,
|
| 1185 |
-
"logits/chosen": -
|
| 1186 |
-
"logits/rejected": -
|
| 1187 |
-
"logps/chosen": -
|
| 1188 |
-
"logps/rejected": -
|
| 1189 |
-
"loss": 0.
|
| 1190 |
-
"rewards/accuracies": 0.
|
| 1191 |
-
"rewards/chosen": -3.
|
| 1192 |
-
"rewards/margins":
|
| 1193 |
-
"rewards/rejected": -5.
|
| 1194 |
"step": 390
|
| 1195 |
},
|
| 1196 |
{
|
| 1197 |
-
"epoch": 0.
|
| 1198 |
-
"grad_norm":
|
| 1199 |
"learning_rate": 4.0416272003232526e-08,
|
| 1200 |
-
"logits/chosen": -
|
| 1201 |
-
"logits/rejected": -
|
| 1202 |
-
"logps/chosen": -
|
| 1203 |
-
"logps/rejected": -
|
| 1204 |
-
"loss": 0.
|
| 1205 |
"rewards/accuracies": 0.856249988079071,
|
| 1206 |
-
"rewards/chosen": -3.
|
| 1207 |
-
"rewards/margins": 2.
|
| 1208 |
-
"rewards/rejected": -5.
|
| 1209 |
"step": 395
|
| 1210 |
},
|
| 1211 |
{
|
| 1212 |
-
"epoch": 0.
|
| 1213 |
-
"grad_norm":
|
| 1214 |
"learning_rate": 3.552332632729041e-08,
|
| 1215 |
-
"logits/chosen": -
|
| 1216 |
-
"logits/rejected": -
|
| 1217 |
-
"logps/chosen": -
|
| 1218 |
-
"logps/rejected": -
|
| 1219 |
-
"loss": 0.
|
| 1220 |
-
"rewards/accuracies": 0.
|
| 1221 |
-
"rewards/chosen": -3.
|
| 1222 |
-
"rewards/margins": 1.
|
| 1223 |
-
"rewards/rejected": -
|
| 1224 |
"step": 400
|
| 1225 |
},
|
| 1226 |
{
|
| 1227 |
-
"epoch": 0.
|
| 1228 |
-
"eval_logits/chosen": -
|
| 1229 |
-
"eval_logits/rejected": -
|
| 1230 |
-
"eval_logps/chosen": -
|
| 1231 |
-
"eval_logps/rejected": -
|
| 1232 |
-
"eval_loss": 0.
|
| 1233 |
-
"eval_rewards/accuracies": 0.
|
| 1234 |
-
"eval_rewards/chosen": -3.
|
| 1235 |
-
"eval_rewards/margins": 1.
|
| 1236 |
-
"eval_rewards/rejected": -5.
|
| 1237 |
-
"eval_runtime":
|
| 1238 |
-
"eval_samples_per_second": 6.
|
| 1239 |
-
"eval_steps_per_second":
|
| 1240 |
"step": 400
|
| 1241 |
},
|
| 1242 |
{
|
| 1243 |
-
"epoch": 0.
|
| 1244 |
-
"grad_norm":
|
| 1245 |
"learning_rate": 3.092332998903416e-08,
|
| 1246 |
-
"logits/chosen": -
|
| 1247 |
-
"logits/rejected": -
|
| 1248 |
-
"logps/chosen": -
|
| 1249 |
-
"logps/rejected": -
|
| 1250 |
-
"loss": 0.
|
| 1251 |
-
"rewards/accuracies": 0.
|
| 1252 |
-
"rewards/chosen": -3.
|
| 1253 |
-
"rewards/margins":
|
| 1254 |
-
"rewards/rejected": -5.
|
| 1255 |
"step": 405
|
| 1256 |
},
|
| 1257 |
{
|
| 1258 |
-
"epoch": 0.
|
| 1259 |
-
"grad_norm":
|
| 1260 |
"learning_rate": 2.6622566030146455e-08,
|
| 1261 |
-
"logits/chosen": -
|
| 1262 |
-
"logits/rejected": -
|
| 1263 |
-
"logps/chosen": -
|
| 1264 |
-
"logps/rejected": -
|
| 1265 |
-
"loss": 0.
|
| 1266 |
-
"rewards/accuracies": 0.
|
| 1267 |
-
"rewards/chosen": -
|
| 1268 |
-
"rewards/margins": 1.
|
| 1269 |
-
"rewards/rejected": -4.
|
| 1270 |
"step": 410
|
| 1271 |
},
|
| 1272 |
{
|
| 1273 |
-
"epoch": 0.
|
| 1274 |
-
"grad_norm":
|
| 1275 |
"learning_rate": 2.26269087768734e-08,
|
| 1276 |
-
"logits/chosen": -
|
| 1277 |
-
"logits/rejected": -
|
| 1278 |
-
"logps/chosen": -
|
| 1279 |
-
"logps/rejected": -
|
| 1280 |
-
"loss": 0.
|
| 1281 |
-
"rewards/accuracies": 0.
|
| 1282 |
-
"rewards/chosen": -3.
|
| 1283 |
-
"rewards/margins": 2.
|
| 1284 |
-
"rewards/rejected": -5.
|
| 1285 |
"step": 415
|
| 1286 |
},
|
| 1287 |
{
|
| 1288 |
-
"epoch": 0.
|
| 1289 |
-
"grad_norm":
|
| 1290 |
"learning_rate": 1.894181581640106e-08,
|
| 1291 |
-
"logits/chosen": -
|
| 1292 |
-
"logits/rejected": -
|
| 1293 |
-
"logps/chosen": -
|
| 1294 |
-
"logps/rejected": -
|
| 1295 |
-
"loss": 0.
|
| 1296 |
-
"rewards/accuracies": 0.
|
| 1297 |
-
"rewards/chosen": -3.
|
| 1298 |
-
"rewards/margins": 1.
|
| 1299 |
-
"rewards/rejected": -
|
| 1300 |
"step": 420
|
| 1301 |
},
|
| 1302 |
{
|
| 1303 |
-
"epoch": 0.
|
| 1304 |
-
"grad_norm":
|
| 1305 |
"learning_rate": 1.5572320542448143e-08,
|
| 1306 |
-
"logits/chosen": -
|
| 1307 |
-
"logits/rejected": -
|
| 1308 |
-
"logps/chosen": -
|
| 1309 |
-
"logps/rejected": -
|
| 1310 |
-
"loss": 0.
|
| 1311 |
-
"rewards/accuracies": 0.
|
| 1312 |
-
"rewards/chosen": -3.
|
| 1313 |
-
"rewards/margins":
|
| 1314 |
-
"rewards/rejected": -5.
|
| 1315 |
"step": 425
|
| 1316 |
},
|
| 1317 |
{
|
| 1318 |
-
"epoch": 0.
|
| 1319 |
-
"grad_norm":
|
| 1320 |
"learning_rate": 1.2523025280255729e-08,
|
| 1321 |
-
"logits/chosen": -
|
| 1322 |
-
"logits/rejected": -
|
| 1323 |
-
"logps/chosen": -
|
| 1324 |
-
"logps/rejected": -
|
| 1325 |
-
"loss": 0.
|
| 1326 |
-
"rewards/accuracies": 0.
|
| 1327 |
-
"rewards/chosen": -3.
|
| 1328 |
-
"rewards/margins":
|
| 1329 |
-
"rewards/rejected": -5.
|
| 1330 |
"step": 430
|
| 1331 |
},
|
| 1332 |
{
|
| 1333 |
-
"epoch": 0.
|
| 1334 |
-
"grad_norm":
|
| 1335 |
"learning_rate": 9.798095000364214e-09,
|
| 1336 |
-
"logits/chosen": -
|
| 1337 |
-
"logits/rejected": -
|
| 1338 |
-
"logps/chosen": -
|
| 1339 |
-
"logps/rejected": -
|
| 1340 |
-
"loss": 0.
|
| 1341 |
-
"rewards/accuracies": 0.
|
| 1342 |
-
"rewards/chosen": -3.
|
| 1343 |
-
"rewards/margins": 2.
|
| 1344 |
-
"rewards/rejected": -5.
|
| 1345 |
"step": 435
|
| 1346 |
},
|
| 1347 |
{
|
| 1348 |
-
"epoch": 0.
|
| 1349 |
-
"grad_norm":
|
| 1350 |
"learning_rate": 7.401251629764876e-09,
|
| 1351 |
-
"logits/chosen": -
|
| 1352 |
-
"logits/rejected": -
|
| 1353 |
-
"logps/chosen": -
|
| 1354 |
-
"logps/rejected": -
|
| 1355 |
-
"loss": 0.
|
| 1356 |
-
"rewards/accuracies": 0.
|
| 1357 |
-
"rewards/chosen": -3.
|
| 1358 |
-
"rewards/margins":
|
| 1359 |
-
"rewards/rejected": -
|
| 1360 |
"step": 440
|
| 1361 |
},
|
| 1362 |
{
|
| 1363 |
-
"epoch": 0.
|
| 1364 |
-
"grad_norm": 18.
|
| 1365 |
"learning_rate": 5.335768968195098e-09,
|
| 1366 |
-
"logits/chosen": -
|
| 1367 |
-
"logits/rejected": -
|
| 1368 |
-
"logps/chosen": -
|
| 1369 |
-
"logps/rejected": -
|
| 1370 |
-
"loss": 0.
|
| 1371 |
-
"rewards/accuracies": 0.
|
| 1372 |
-
"rewards/chosen": -3.
|
| 1373 |
-
"rewards/margins":
|
| 1374 |
-
"rewards/rejected": -5.
|
| 1375 |
"step": 445
|
| 1376 |
},
|
| 1377 |
{
|
| 1378 |
-
"epoch": 0.
|
| 1379 |
-
"grad_norm":
|
| 1380 |
"learning_rate": 3.604468216521883e-09,
|
| 1381 |
-
"logits/chosen": -
|
| 1382 |
-
"logits/rejected": -
|
| 1383 |
-
"logps/chosen": -
|
| 1384 |
-
"logps/rejected": -
|
| 1385 |
-
"loss": 0.
|
| 1386 |
-
"rewards/accuracies": 0.
|
| 1387 |
-
"rewards/chosen": -3.
|
| 1388 |
-
"rewards/margins": 1.
|
| 1389 |
-
"rewards/rejected": -5.
|
| 1390 |
"step": 450
|
| 1391 |
},
|
| 1392 |
{
|
| 1393 |
-
"epoch": 0.
|
| 1394 |
-
"grad_norm":
|
| 1395 |
"learning_rate": 2.2097141233206884e-09,
|
| 1396 |
-
"logits/chosen": -
|
| 1397 |
-
"logits/rejected": -
|
| 1398 |
-
"logps/chosen": -
|
| 1399 |
-
"logps/rejected": -
|
| 1400 |
-
"loss": 0.
|
| 1401 |
-
"rewards/accuracies": 0.
|
| 1402 |
-
"rewards/chosen": -3.
|
| 1403 |
-
"rewards/margins":
|
| 1404 |
-
"rewards/rejected": -5.
|
| 1405 |
"step": 455
|
| 1406 |
},
|
| 1407 |
{
|
| 1408 |
-
"epoch": 0.
|
| 1409 |
-
"grad_norm":
|
| 1410 |
"learning_rate": 1.1534117549133472e-09,
|
| 1411 |
-
"logits/chosen": -
|
| 1412 |
-
"logits/rejected": -
|
| 1413 |
-
"logps/chosen": -
|
| 1414 |
-
"logps/rejected": -
|
| 1415 |
-
"loss": 0.
|
| 1416 |
-
"rewards/accuracies": 0.
|
| 1417 |
-
"rewards/chosen": -3.
|
| 1418 |
-
"rewards/margins": 2.
|
| 1419 |
-
"rewards/rejected": -5.
|
| 1420 |
"step": 460
|
| 1421 |
},
|
| 1422 |
{
|
| 1423 |
-
"epoch": 0.
|
| 1424 |
-
"grad_norm":
|
| 1425 |
"learning_rate": 4.3700389327672173e-10,
|
| 1426 |
-
"logits/chosen": -
|
| 1427 |
-
"logits/rejected": -
|
| 1428 |
-
"logps/chosen": -
|
| 1429 |
-
"logps/rejected": -
|
| 1430 |
-
"loss": 0.
|
| 1431 |
-
"rewards/accuracies": 0.
|
| 1432 |
-
"rewards/chosen": -3.
|
| 1433 |
-
"rewards/margins":
|
| 1434 |
-
"rewards/rejected": -5.
|
| 1435 |
"step": 465
|
| 1436 |
},
|
| 1437 |
{
|
| 1438 |
-
"epoch": 0.
|
| 1439 |
-
"grad_norm":
|
| 1440 |
"learning_rate": 6.146906537587982e-11,
|
| 1441 |
-
"logits/chosen": -
|
| 1442 |
-
"logits/rejected": -
|
| 1443 |
-
"logps/chosen": -600.
|
| 1444 |
-
"logps/rejected": -
|
| 1445 |
-
"loss": 0.
|
| 1446 |
-
"rewards/accuracies": 0.
|
| 1447 |
-
"rewards/chosen": -3.
|
| 1448 |
-
"rewards/margins":
|
| 1449 |
-
"rewards/rejected": -5.
|
| 1450 |
"step": 470
|
| 1451 |
},
|
| 1452 |
{
|
| 1453 |
-
"epoch": 0.
|
| 1454 |
"step": 473,
|
| 1455 |
"total_flos": 0.0,
|
| 1456 |
-
"train_loss": 0.
|
| 1457 |
-
"train_runtime":
|
| 1458 |
-
"train_samples_per_second":
|
| 1459 |
-
"train_steps_per_second": 0.
|
| 1460 |
}
|
| 1461 |
],
|
| 1462 |
"logging_steps": 5,
|
|
@@ -1477,7 +1477,7 @@
|
|
| 1477 |
}
|
| 1478 |
},
|
| 1479 |
"total_flos": 0.0,
|
| 1480 |
-
"train_batch_size":
|
| 1481 |
"trial_name": null,
|
| 1482 |
"trial_params": null
|
| 1483 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9984827495217362,
|
| 5 |
"eval_steps": 400,
|
| 6 |
"global_step": 473,
|
| 7 |
"is_hyper_param_search": false,
|
|
|
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
+
"epoch": 0.002110957187149548,
|
| 13 |
+
"grad_norm": 4.06041781261902,
|
| 14 |
"learning_rate": 1.0416666666666666e-08,
|
| 15 |
+
"logits/chosen": -0.9878771901130676,
|
| 16 |
+
"logits/rejected": -0.7230668663978577,
|
| 17 |
+
"logps/chosen": -251.34963989257812,
|
| 18 |
+
"logps/rejected": -287.15838623046875,
|
| 19 |
"loss": 0.6931,
|
| 20 |
"rewards/accuracies": 0.0,
|
| 21 |
"rewards/chosen": 0.0,
|
|
|
|
| 24 |
"step": 1
|
| 25 |
},
|
| 26 |
{
|
| 27 |
+
"epoch": 0.01055478593574774,
|
| 28 |
+
"grad_norm": 4.105776204209711,
|
| 29 |
"learning_rate": 5.208333333333333e-08,
|
| 30 |
+
"logits/chosen": -0.9975427985191345,
|
| 31 |
+
"logits/rejected": -0.6888133883476257,
|
| 32 |
+
"logps/chosen": -272.86993408203125,
|
| 33 |
+
"logps/rejected": -286.1126708984375,
|
| 34 |
+
"loss": 0.6932,
|
| 35 |
+
"rewards/accuracies": 0.359375,
|
| 36 |
+
"rewards/chosen": -0.00011302110215183347,
|
| 37 |
+
"rewards/margins": -0.00016400158347096294,
|
| 38 |
+
"rewards/rejected": 5.0980423111468554e-05,
|
| 39 |
"step": 5
|
| 40 |
},
|
| 41 |
{
|
| 42 |
+
"epoch": 0.02110957187149548,
|
| 43 |
+
"grad_norm": 4.582740122598074,
|
| 44 |
"learning_rate": 1.0416666666666667e-07,
|
| 45 |
+
"logits/chosen": -1.0306963920593262,
|
| 46 |
+
"logits/rejected": -0.7614760398864746,
|
| 47 |
+
"logps/chosen": -274.9240417480469,
|
| 48 |
+
"logps/rejected": -293.0102233886719,
|
| 49 |
"loss": 0.6933,
|
| 50 |
+
"rewards/accuracies": 0.5625,
|
| 51 |
+
"rewards/chosen": 0.0007906880346126854,
|
| 52 |
+
"rewards/margins": 0.00033544833422638476,
|
| 53 |
+
"rewards/rejected": 0.00045523978769779205,
|
| 54 |
"step": 10
|
| 55 |
},
|
| 56 |
{
|
| 57 |
+
"epoch": 0.03166435780724322,
|
| 58 |
+
"grad_norm": 4.2774752538157,
|
| 59 |
"learning_rate": 1.5624999999999999e-07,
|
| 60 |
+
"logits/chosen": -0.9785528182983398,
|
| 61 |
+
"logits/rejected": -0.6860870122909546,
|
| 62 |
+
"logps/chosen": -253.0674285888672,
|
| 63 |
+
"logps/rejected": -277.63006591796875,
|
| 64 |
+
"loss": 0.693,
|
| 65 |
+
"rewards/accuracies": 0.44999998807907104,
|
| 66 |
+
"rewards/chosen": 0.00013054809824097902,
|
| 67 |
+
"rewards/margins": -4.017539322376251e-05,
|
| 68 |
+
"rewards/rejected": 0.0001707235351204872,
|
| 69 |
"step": 15
|
| 70 |
},
|
| 71 |
{
|
| 72 |
+
"epoch": 0.04221914374299096,
|
| 73 |
+
"grad_norm": 4.154426876518502,
|
| 74 |
"learning_rate": 2.0833333333333333e-07,
|
| 75 |
+
"logits/chosen": -0.8861902356147766,
|
| 76 |
+
"logits/rejected": -0.7466350793838501,
|
| 77 |
+
"logps/chosen": -303.89990234375,
|
| 78 |
+
"logps/rejected": -317.44354248046875,
|
| 79 |
+
"loss": 0.6929,
|
| 80 |
+
"rewards/accuracies": 0.48750001192092896,
|
| 81 |
+
"rewards/chosen": -0.0017729544779285789,
|
| 82 |
+
"rewards/margins": 0.0007360944291576743,
|
| 83 |
+
"rewards/rejected": -0.0025090486742556095,
|
| 84 |
"step": 20
|
| 85 |
},
|
| 86 |
{
|
| 87 |
+
"epoch": 0.0527739296787387,
|
| 88 |
+
"grad_norm": 4.391172268221911,
|
| 89 |
"learning_rate": 2.604166666666667e-07,
|
| 90 |
+
"logits/chosen": -0.9347459077835083,
|
| 91 |
+
"logits/rejected": -0.7482324242591858,
|
| 92 |
+
"logps/chosen": -257.277099609375,
|
| 93 |
+
"logps/rejected": -276.6146240234375,
|
| 94 |
+
"loss": 0.6925,
|
| 95 |
+
"rewards/accuracies": 0.5625,
|
| 96 |
+
"rewards/chosen": -0.0021017056424170732,
|
| 97 |
+
"rewards/margins": 0.001351111801341176,
|
| 98 |
+
"rewards/rejected": -0.003452816978096962,
|
| 99 |
"step": 25
|
| 100 |
},
|
| 101 |
{
|
| 102 |
+
"epoch": 0.06332871561448644,
|
| 103 |
+
"grad_norm": 4.197922988205761,
|
| 104 |
"learning_rate": 3.1249999999999997e-07,
|
| 105 |
+
"logits/chosen": -0.9495643377304077,
|
| 106 |
+
"logits/rejected": -0.722493052482605,
|
| 107 |
+
"logps/chosen": -271.043212890625,
|
| 108 |
+
"logps/rejected": -304.54779052734375,
|
| 109 |
+
"loss": 0.6909,
|
| 110 |
+
"rewards/accuracies": 0.675000011920929,
|
| 111 |
+
"rewards/chosen": -0.007976134307682514,
|
| 112 |
+
"rewards/margins": 0.006574218161404133,
|
| 113 |
+
"rewards/rejected": -0.014550352469086647,
|
| 114 |
"step": 30
|
| 115 |
},
|
| 116 |
{
|
| 117 |
+
"epoch": 0.07388350155023418,
|
| 118 |
+
"grad_norm": 4.146928571575294,
|
| 119 |
"learning_rate": 3.645833333333333e-07,
|
| 120 |
+
"logits/chosen": -0.9740797281265259,
|
| 121 |
+
"logits/rejected": -0.6511734127998352,
|
| 122 |
+
"logps/chosen": -269.66558837890625,
|
| 123 |
+
"logps/rejected": -282.5665588378906,
|
| 124 |
+
"loss": 0.6901,
|
| 125 |
+
"rewards/accuracies": 0.643750011920929,
|
| 126 |
+
"rewards/chosen": -0.014239413663744926,
|
| 127 |
+
"rewards/margins": 0.006516980938613415,
|
| 128 |
+
"rewards/rejected": -0.020756395533680916,
|
| 129 |
"step": 35
|
| 130 |
},
|
| 131 |
{
|
| 132 |
+
"epoch": 0.08443828748598192,
|
| 133 |
+
"grad_norm": 4.434625304551974,
|
| 134 |
"learning_rate": 4.1666666666666667e-07,
|
| 135 |
+
"logits/chosen": -0.9476411938667297,
|
| 136 |
+
"logits/rejected": -0.8075205087661743,
|
| 137 |
+
"logps/chosen": -254.89315795898438,
|
| 138 |
+
"logps/rejected": -269.08843994140625,
|
| 139 |
+
"loss": 0.6872,
|
| 140 |
+
"rewards/accuracies": 0.706250011920929,
|
| 141 |
+
"rewards/chosen": -0.025963688269257545,
|
| 142 |
+
"rewards/margins": 0.010111861862242222,
|
| 143 |
+
"rewards/rejected": -0.03607555106282234,
|
| 144 |
"step": 40
|
| 145 |
},
|
| 146 |
{
|
| 147 |
+
"epoch": 0.09499307342172966,
|
| 148 |
+
"grad_norm": 4.334333725960549,
|
| 149 |
"learning_rate": 4.6874999999999996e-07,
|
| 150 |
+
"logits/chosen": -0.9668118357658386,
|
| 151 |
+
"logits/rejected": -0.7818160057067871,
|
| 152 |
+
"logps/chosen": -262.55804443359375,
|
| 153 |
+
"logps/rejected": -279.33416748046875,
|
| 154 |
+
"loss": 0.6841,
|
| 155 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 156 |
+
"rewards/chosen": -0.03984779864549637,
|
| 157 |
+
"rewards/margins": 0.01882680132985115,
|
| 158 |
+
"rewards/rejected": -0.05867459625005722,
|
| 159 |
"step": 45
|
| 160 |
},
|
| 161 |
{
|
| 162 |
+
"epoch": 0.1055478593574774,
|
| 163 |
+
"grad_norm": 4.672589574927753,
|
| 164 |
"learning_rate": 4.999726797933858e-07,
|
| 165 |
+
"logits/chosen": -1.1188008785247803,
|
| 166 |
+
"logits/rejected": -0.7495776414871216,
|
| 167 |
+
"logps/chosen": -274.8739318847656,
|
| 168 |
+
"logps/rejected": -298.7273254394531,
|
| 169 |
+
"loss": 0.6736,
|
| 170 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 171 |
+
"rewards/chosen": -0.0697299912571907,
|
| 172 |
+
"rewards/margins": 0.04492691531777382,
|
| 173 |
+
"rewards/rejected": -0.11465690284967422,
|
| 174 |
"step": 50
|
| 175 |
},
|
| 176 |
{
|
| 177 |
+
"epoch": 0.11610264529322514,
|
| 178 |
+
"grad_norm": 6.181545467894853,
|
| 179 |
"learning_rate": 4.99665396039775e-07,
|
| 180 |
+
"logits/chosen": -1.1051629781723022,
|
| 181 |
+
"logits/rejected": -0.9252668619155884,
|
| 182 |
+
"logps/chosen": -270.87841796875,
|
| 183 |
+
"logps/rejected": -289.1105651855469,
|
| 184 |
+
"loss": 0.6576,
|
| 185 |
+
"rewards/accuracies": 0.731249988079071,
|
| 186 |
+
"rewards/chosen": -0.14251390099525452,
|
| 187 |
+
"rewards/margins": 0.10730250179767609,
|
| 188 |
+
"rewards/rejected": -0.2498163878917694,
|
| 189 |
"step": 55
|
| 190 |
},
|
| 191 |
{
|
| 192 |
+
"epoch": 0.12665743122897288,
|
| 193 |
+
"grad_norm": 7.558458036407823,
|
| 194 |
"learning_rate": 4.99017099386437e-07,
|
| 195 |
+
"logits/chosen": -1.3753994703292847,
|
| 196 |
+
"logits/rejected": -1.1374595165252686,
|
| 197 |
+
"logps/chosen": -306.73138427734375,
|
| 198 |
+
"logps/rejected": -341.1016845703125,
|
| 199 |
+
"loss": 0.6281,
|
| 200 |
"rewards/accuracies": 0.75,
|
| 201 |
+
"rewards/chosen": -0.37672942876815796,
|
| 202 |
+
"rewards/margins": 0.20342817902565002,
|
| 203 |
+
"rewards/rejected": -0.5801576375961304,
|
| 204 |
"step": 60
|
| 205 |
},
|
| 206 |
{
|
| 207 |
+
"epoch": 0.13721221716472062,
|
| 208 |
+
"grad_norm": 66.9078697860869,
|
| 209 |
"learning_rate": 4.980286753286194e-07,
|
| 210 |
+
"logits/chosen": -1.5234780311584473,
|
| 211 |
+
"logits/rejected": -1.2897632122039795,
|
| 212 |
+
"logps/chosen": -359.58074951171875,
|
| 213 |
+
"logps/rejected": -414.0733947753906,
|
| 214 |
+
"loss": 0.6211,
|
| 215 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 216 |
+
"rewards/chosen": -1.063786506652832,
|
| 217 |
+
"rewards/margins": 0.41773301362991333,
|
| 218 |
+
"rewards/rejected": -1.4815195798873901,
|
| 219 |
"step": 65
|
| 220 |
},
|
| 221 |
{
|
| 222 |
+
"epoch": 0.14776700310046836,
|
| 223 |
+
"grad_norm": 12.864645326069963,
|
| 224 |
"learning_rate": 4.967014739346915e-07,
|
| 225 |
+
"logits/chosen": -1.4892756938934326,
|
| 226 |
+
"logits/rejected": -1.3145514726638794,
|
| 227 |
+
"logps/chosen": -410.1991271972656,
|
| 228 |
+
"logps/rejected": -467.57623291015625,
|
| 229 |
+
"loss": 0.6578,
|
| 230 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 231 |
+
"rewards/chosen": -1.4497530460357666,
|
| 232 |
+
"rewards/margins": 0.4113241136074066,
|
| 233 |
+
"rewards/rejected": -1.8610769510269165,
|
| 234 |
"step": 70
|
| 235 |
},
|
| 236 |
{
|
| 237 |
+
"epoch": 0.1583217890362161,
|
| 238 |
+
"grad_norm": 12.52823747211519,
|
| 239 |
"learning_rate": 4.950373080021136e-07,
|
| 240 |
+
"logits/chosen": -1.2871811389923096,
|
| 241 |
+
"logits/rejected": -1.2107694149017334,
|
| 242 |
+
"logps/chosen": -333.7770080566406,
|
| 243 |
+
"logps/rejected": -356.77728271484375,
|
| 244 |
+
"loss": 0.5936,
|
| 245 |
+
"rewards/accuracies": 0.6875,
|
| 246 |
+
"rewards/chosen": -0.579529881477356,
|
| 247 |
+
"rewards/margins": 0.2453722506761551,
|
| 248 |
+
"rewards/rejected": -0.8249020576477051,
|
| 249 |
"step": 75
|
| 250 |
},
|
| 251 |
{
|
| 252 |
+
"epoch": 0.16887657497196384,
|
| 253 |
+
"grad_norm": 17.45417402457315,
|
| 254 |
"learning_rate": 4.930384505813737e-07,
|
| 255 |
+
"logits/chosen": -1.4544992446899414,
|
| 256 |
+
"logits/rejected": -1.3903002738952637,
|
| 257 |
+
"logps/chosen": -392.7482604980469,
|
| 258 |
+
"logps/rejected": -472.0469665527344,
|
| 259 |
+
"loss": 0.5588,
|
| 260 |
+
"rewards/accuracies": 0.762499988079071,
|
| 261 |
+
"rewards/chosen": -1.0880345106124878,
|
| 262 |
+
"rewards/margins": 0.6102110147476196,
|
| 263 |
+
"rewards/rejected": -1.6982454061508179,
|
| 264 |
"step": 80
|
| 265 |
},
|
| 266 |
{
|
| 267 |
+
"epoch": 0.17943136090771158,
|
| 268 |
+
"grad_norm": 36.09231440231977,
|
| 269 |
"learning_rate": 4.907076318712738e-07,
|
| 270 |
+
"logits/chosen": -1.5770976543426514,
|
| 271 |
+
"logits/rejected": -1.4604318141937256,
|
| 272 |
+
"logps/chosen": -478.90264892578125,
|
| 273 |
+
"logps/rejected": -560.0563354492188,
|
| 274 |
+
"loss": 0.557,
|
| 275 |
+
"rewards/accuracies": 0.699999988079071,
|
| 276 |
+
"rewards/chosen": -1.8632774353027344,
|
| 277 |
+
"rewards/margins": 0.6757477521896362,
|
| 278 |
+
"rewards/rejected": -2.53902530670166,
|
| 279 |
"step": 85
|
| 280 |
},
|
| 281 |
{
|
| 282 |
+
"epoch": 0.18998614684345932,
|
| 283 |
+
"grad_norm": 12.93821128098639,
|
| 284 |
"learning_rate": 4.88048035489807e-07,
|
| 285 |
+
"logits/chosen": -1.5192902088165283,
|
| 286 |
+
"logits/rejected": -1.3578670024871826,
|
| 287 |
+
"logps/chosen": -401.8139343261719,
|
| 288 |
+
"logps/rejected": -476.27557373046875,
|
| 289 |
+
"loss": 0.5413,
|
| 290 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 291 |
+
"rewards/chosen": -1.2336255311965942,
|
| 292 |
+
"rewards/margins": 0.6196562051773071,
|
| 293 |
+
"rewards/rejected": -1.8532816171646118,
|
| 294 |
"step": 90
|
| 295 |
},
|
| 296 |
{
|
| 297 |
+
"epoch": 0.20054093277920707,
|
| 298 |
+
"grad_norm": 10.174846260495245,
|
| 299 |
"learning_rate": 4.85063294125718e-07,
|
| 300 |
+
"logits/chosen": -1.4184257984161377,
|
| 301 |
+
"logits/rejected": -1.349448561668396,
|
| 302 |
+
"logps/chosen": -405.37030029296875,
|
| 303 |
+
"logps/rejected": -465.87835693359375,
|
| 304 |
+
"loss": 0.584,
|
| 305 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 306 |
+
"rewards/chosen": -1.2496789693832397,
|
| 307 |
+
"rewards/margins": 0.4809112548828125,
|
| 308 |
+
"rewards/rejected": -1.7305902242660522,
|
| 309 |
"step": 95
|
| 310 |
},
|
| 311 |
{
|
| 312 |
+
"epoch": 0.2110957187149548,
|
| 313 |
+
"grad_norm": 11.934921621793764,
|
| 314 |
"learning_rate": 4.817574845766874e-07,
|
| 315 |
+
"logits/chosen": -1.6645218133926392,
|
| 316 |
+
"logits/rejected": -1.4362655878067017,
|
| 317 |
+
"logps/chosen": -419.6376953125,
|
| 318 |
+
"logps/rejected": -511.97955322265625,
|
| 319 |
+
"loss": 0.5179,
|
| 320 |
"rewards/accuracies": 0.7749999761581421,
|
| 321 |
+
"rewards/chosen": -1.6008678674697876,
|
| 322 |
+
"rewards/margins": 0.7449867129325867,
|
| 323 |
+
"rewards/rejected": -2.3458542823791504,
|
| 324 |
"step": 100
|
| 325 |
},
|
| 326 |
{
|
| 327 |
+
"epoch": 0.22165050465070255,
|
| 328 |
+
"grad_norm": 11.911318908319176,
|
| 329 |
"learning_rate": 4.781351221809166e-07,
|
| 330 |
+
"logits/chosen": -1.6439392566680908,
|
| 331 |
+
"logits/rejected": -1.5432502031326294,
|
| 332 |
+
"logps/chosen": -462.12249755859375,
|
| 333 |
+
"logps/rejected": -561.0224609375,
|
| 334 |
+
"loss": 0.563,
|
| 335 |
+
"rewards/accuracies": 0.737500011920929,
|
| 336 |
+
"rewards/chosen": -2.0325913429260254,
|
| 337 |
+
"rewards/margins": 0.8397369384765625,
|
| 338 |
+
"rewards/rejected": -2.872328281402588,
|
| 339 |
"step": 105
|
| 340 |
},
|
| 341 |
{
|
| 342 |
+
"epoch": 0.23220529058645029,
|
| 343 |
+
"grad_norm": 12.017204037189543,
|
| 344 |
"learning_rate": 4.742011546497182e-07,
|
| 345 |
+
"logits/chosen": -1.5051389932632446,
|
| 346 |
+
"logits/rejected": -1.4414231777191162,
|
| 347 |
+
"logps/chosen": -449.30657958984375,
|
| 348 |
+
"logps/rejected": -550.5023803710938,
|
| 349 |
+
"loss": 0.5068,
|
| 350 |
+
"rewards/accuracies": 0.8062499761581421,
|
| 351 |
+
"rewards/chosen": -1.5960687398910522,
|
| 352 |
+
"rewards/margins": 0.8259360194206238,
|
| 353 |
+
"rewards/rejected": -2.4220046997070312,
|
| 354 |
"step": 110
|
| 355 |
},
|
| 356 |
{
|
| 357 |
+
"epoch": 0.24276007652219803,
|
| 358 |
+
"grad_norm": 16.38172474746581,
|
| 359 |
"learning_rate": 4.6996095530953875e-07,
|
| 360 |
+
"logits/chosen": -1.6456964015960693,
|
| 361 |
+
"logits/rejected": -1.4991674423217773,
|
| 362 |
+
"logps/chosen": -492.7826232910156,
|
| 363 |
+
"logps/rejected": -631.8049926757812,
|
| 364 |
+
"loss": 0.5084,
|
| 365 |
+
"rewards/accuracies": 0.824999988079071,
|
| 366 |
+
"rewards/chosen": -2.1061949729919434,
|
| 367 |
+
"rewards/margins": 1.1719015836715698,
|
| 368 |
+
"rewards/rejected": -3.2780966758728027,
|
| 369 |
"step": 115
|
| 370 |
},
|
| 371 |
{
|
| 372 |
+
"epoch": 0.25331486245794577,
|
| 373 |
+
"grad_norm": 19.153944069009537,
|
| 374 |
"learning_rate": 4.654203157626399e-07,
|
| 375 |
+
"logits/chosen": -1.7168292999267578,
|
| 376 |
+
"logits/rejected": -1.5979254245758057,
|
| 377 |
+
"logps/chosen": -499.25775146484375,
|
| 378 |
+
"logps/rejected": -663.107177734375,
|
| 379 |
+
"loss": 0.5023,
|
| 380 |
+
"rewards/accuracies": 0.762499988079071,
|
| 381 |
+
"rewards/chosen": -2.420952558517456,
|
| 382 |
+
"rewards/margins": 1.3582748174667358,
|
| 383 |
+
"rewards/rejected": -3.7792270183563232,
|
| 384 |
"step": 120
|
| 385 |
},
|
| 386 |
{
|
| 387 |
+
"epoch": 0.26386964839369353,
|
| 388 |
+
"grad_norm": 15.209663016177657,
|
| 389 |
"learning_rate": 4.605854379764673e-07,
|
| 390 |
+
"logits/chosen": -1.529827356338501,
|
| 391 |
+
"logits/rejected": -1.4732040166854858,
|
| 392 |
+
"logps/chosen": -422.29827880859375,
|
| 393 |
+
"logps/rejected": -505.83154296875,
|
| 394 |
+
"loss": 0.4999,
|
| 395 |
+
"rewards/accuracies": 0.800000011920929,
|
| 396 |
+
"rewards/chosen": -1.6309483051300049,
|
| 397 |
+
"rewards/margins": 0.74617999792099,
|
| 398 |
+
"rewards/rejected": -2.3771283626556396,
|
| 399 |
"step": 125
|
| 400 |
},
|
| 401 |
{
|
| 402 |
+
"epoch": 0.27442443432944125,
|
| 403 |
+
"grad_norm": 11.779884218050361,
|
| 404 |
"learning_rate": 4.5546292581250857e-07,
|
| 405 |
+
"logits/chosen": -1.5948470830917358,
|
| 406 |
+
"logits/rejected": -1.4933321475982666,
|
| 407 |
+
"logps/chosen": -455.269287109375,
|
| 408 |
+
"logps/rejected": -543.4944458007812,
|
| 409 |
+
"loss": 0.4868,
|
| 410 |
+
"rewards/accuracies": 0.8062499761581421,
|
| 411 |
+
"rewards/chosen": -1.7508251667022705,
|
| 412 |
+
"rewards/margins": 0.8269698023796082,
|
| 413 |
+
"rewards/rejected": -2.5777950286865234,
|
| 414 |
"step": 130
|
| 415 |
},
|
| 416 |
{
|
| 417 |
+
"epoch": 0.284979220265189,
|
| 418 |
+
"grad_norm": 19.10994211444829,
|
| 419 |
"learning_rate": 4.5005977600621275e-07,
|
| 420 |
+
"logits/chosen": -1.7589390277862549,
|
| 421 |
+
"logits/rejected": -1.5872992277145386,
|
| 422 |
+
"logps/chosen": -568.4318237304688,
|
| 423 |
+
"logps/rejected": -727.857666015625,
|
| 424 |
+
"loss": 0.4781,
|
| 425 |
+
"rewards/accuracies": 0.75,
|
| 426 |
+
"rewards/chosen": -2.8650763034820557,
|
| 427 |
+
"rewards/margins": 1.3692649602890015,
|
| 428 |
+
"rewards/rejected": -4.234341621398926,
|
| 429 |
"step": 135
|
| 430 |
},
|
| 431 |
{
|
| 432 |
+
"epoch": 0.2955340062009367,
|
| 433 |
+
"grad_norm": 15.492507693472131,
|
| 434 |
"learning_rate": 4.443833686102919e-07,
|
| 435 |
+
"logits/chosen": -1.6332670450210571,
|
| 436 |
+
"logits/rejected": -1.4034180641174316,
|
| 437 |
+
"logps/chosen": -461.40826416015625,
|
| 438 |
+
"logps/rejected": -608.6381225585938,
|
| 439 |
+
"loss": 0.4705,
|
| 440 |
+
"rewards/accuracies": 0.824999988079071,
|
| 441 |
+
"rewards/chosen": -1.9456875324249268,
|
| 442 |
+
"rewards/margins": 1.1761558055877686,
|
| 443 |
+
"rewards/rejected": -3.1218433380126953,
|
| 444 |
"step": 140
|
| 445 |
},
|
| 446 |
{
|
| 447 |
+
"epoch": 0.3060887921366845,
|
| 448 |
+
"grad_norm": 16.867870056444893,
|
| 449 |
"learning_rate": 4.384414569144561e-07,
|
| 450 |
+
"logits/chosen": -1.5909126996994019,
|
| 451 |
+
"logits/rejected": -1.5102109909057617,
|
| 452 |
+
"logps/chosen": -441.18670654296875,
|
| 453 |
+
"logps/rejected": -565.0006713867188,
|
| 454 |
+
"loss": 0.5053,
|
| 455 |
+
"rewards/accuracies": 0.793749988079071,
|
| 456 |
+
"rewards/chosen": -1.909717321395874,
|
| 457 |
+
"rewards/margins": 1.0274746417999268,
|
| 458 |
+
"rewards/rejected": -2.937191963195801,
|
| 459 |
"step": 145
|
| 460 |
},
|
| 461 |
{
|
| 462 |
+
"epoch": 0.3166435780724322,
|
| 463 |
+
"grad_norm": 12.53246744632567,
|
| 464 |
"learning_rate": 4.3224215685535287e-07,
|
| 465 |
+
"logits/chosen": -1.6234986782073975,
|
| 466 |
+
"logits/rejected": -1.4319360256195068,
|
| 467 |
+
"logps/chosen": -540.7374267578125,
|
| 468 |
+
"logps/rejected": -702.7229614257812,
|
| 469 |
+
"loss": 0.4621,
|
| 470 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 471 |
+
"rewards/chosen": -2.5035765171051025,
|
| 472 |
+
"rewards/margins": 1.4901249408721924,
|
| 473 |
+
"rewards/rejected": -3.993701457977295,
|
| 474 |
"step": 150
|
| 475 |
},
|
| 476 |
{
|
| 477 |
+
"epoch": 0.32719836400818,
|
| 478 |
+
"grad_norm": 15.04185556896974,
|
| 479 |
"learning_rate": 4.2579393593117364e-07,
|
| 480 |
+
"logits/chosen": -1.533140778541565,
|
| 481 |
+
"logits/rejected": -1.3760929107666016,
|
| 482 |
+
"logps/chosen": -481.59954833984375,
|
| 483 |
+
"logps/rejected": -652.427490234375,
|
| 484 |
+
"loss": 0.4563,
|
| 485 |
+
"rewards/accuracies": 0.8187500238418579,
|
| 486 |
+
"rewards/chosen": -2.2860851287841797,
|
| 487 |
+
"rewards/margins": 1.5269566774368286,
|
| 488 |
+
"rewards/rejected": -3.8130416870117188,
|
| 489 |
"step": 155
|
| 490 |
},
|
| 491 |
{
|
| 492 |
+
"epoch": 0.3377531499439277,
|
| 493 |
+
"grad_norm": 13.12290477739756,
|
| 494 |
"learning_rate": 4.191056016360699e-07,
|
| 495 |
+
"logits/chosen": -1.6271283626556396,
|
| 496 |
+
"logits/rejected": -1.5437796115875244,
|
| 497 |
+
"logps/chosen": -519.0302734375,
|
| 498 |
+
"logps/rejected": -663.0972900390625,
|
| 499 |
+
"loss": 0.4474,
|
| 500 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 501 |
+
"rewards/chosen": -2.57169771194458,
|
| 502 |
+
"rewards/margins": 1.3303985595703125,
|
| 503 |
+
"rewards/rejected": -3.9020965099334717,
|
| 504 |
"step": 160
|
| 505 |
},
|
| 506 |
{
|
| 507 |
+
"epoch": 0.34830793587967546,
|
| 508 |
+
"grad_norm": 15.784620667841397,
|
| 509 |
"learning_rate": 4.121862894301754e-07,
|
| 510 |
+
"logits/chosen": -1.5655263662338257,
|
| 511 |
+
"logits/rejected": -1.5315742492675781,
|
| 512 |
+
"logps/chosen": -580.234375,
|
| 513 |
+
"logps/rejected": -725.3365478515625,
|
| 514 |
+
"loss": 0.4634,
|
| 515 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 516 |
+
"rewards/chosen": -2.9336700439453125,
|
| 517 |
+
"rewards/margins": 1.4053281545639038,
|
| 518 |
+
"rewards/rejected": -4.338997840881348,
|
| 519 |
"step": 165
|
| 520 |
},
|
| 521 |
{
|
| 522 |
+
"epoch": 0.35886272181542317,
|
| 523 |
+
"grad_norm": 18.950750373982537,
|
| 524 |
"learning_rate": 4.050454502616667e-07,
|
| 525 |
+
"logits/chosen": -1.7045695781707764,
|
| 526 |
+
"logits/rejected": -1.6517536640167236,
|
| 527 |
+
"logps/chosen": -554.7913208007812,
|
| 528 |
+
"logps/rejected": -716.6380615234375,
|
| 529 |
+
"loss": 0.4487,
|
| 530 |
+
"rewards/accuracies": 0.8062499761581421,
|
| 531 |
+
"rewards/chosen": -2.823065757751465,
|
| 532 |
+
"rewards/margins": 1.4535109996795654,
|
| 533 |
+
"rewards/rejected": -4.276576519012451,
|
| 534 |
"step": 170
|
| 535 |
},
|
| 536 |
{
|
| 537 |
+
"epoch": 0.36941750775117094,
|
| 538 |
+
"grad_norm": 17.91495880964947,
|
| 539 |
"learning_rate": 3.976928376579047e-07,
|
| 540 |
+
"logits/chosen": -1.6950843334197998,
|
| 541 |
+
"logits/rejected": -1.582262396812439,
|
| 542 |
+
"logps/chosen": -520.1227416992188,
|
| 543 |
+
"logps/rejected": -664.7024536132812,
|
| 544 |
+
"loss": 0.445,
|
| 545 |
+
"rewards/accuracies": 0.824999988079071,
|
| 546 |
+
"rewards/chosen": -2.5456645488739014,
|
| 547 |
+
"rewards/margins": 1.2352759838104248,
|
| 548 |
+
"rewards/rejected": -3.780940532684326,
|
| 549 |
"step": 175
|
| 550 |
},
|
| 551 |
{
|
| 552 |
+
"epoch": 0.37997229368691865,
|
| 553 |
+
"grad_norm": 15.17521580280272,
|
| 554 |
"learning_rate": 3.9013849440328945e-07,
|
| 555 |
+
"logits/chosen": -1.6940500736236572,
|
| 556 |
+
"logits/rejected": -1.5467922687530518,
|
| 557 |
+
"logps/chosen": -520.3582763671875,
|
| 558 |
+
"logps/rejected": -680.7635498046875,
|
| 559 |
+
"loss": 0.4665,
|
| 560 |
+
"rewards/accuracies": 0.8125,
|
| 561 |
+
"rewards/chosen": -2.551764726638794,
|
| 562 |
+
"rewards/margins": 1.372143268585205,
|
| 563 |
+
"rewards/rejected": -3.92390775680542,
|
| 564 |
"step": 180
|
| 565 |
},
|
| 566 |
{
|
| 567 |
+
"epoch": 0.3905270796226664,
|
| 568 |
+
"grad_norm": 14.308326454629483,
|
| 569 |
"learning_rate": 3.8239273882202473e-07,
|
| 570 |
+
"logits/chosen": -1.6619869470596313,
|
| 571 |
+
"logits/rejected": -1.5419933795928955,
|
| 572 |
+
"logps/chosen": -506.9136657714844,
|
| 573 |
+
"logps/rejected": -639.9857177734375,
|
| 574 |
+
"loss": 0.4574,
|
| 575 |
+
"rewards/accuracies": 0.737500011920929,
|
| 576 |
+
"rewards/chosen": -2.3956522941589355,
|
| 577 |
+
"rewards/margins": 1.23929762840271,
|
| 578 |
+
"rewards/rejected": -3.6349494457244873,
|
| 579 |
"step": 185
|
| 580 |
},
|
| 581 |
{
|
| 582 |
+
"epoch": 0.40108186555841413,
|
| 583 |
+
"grad_norm": 15.767200038387838,
|
| 584 |
"learning_rate": 3.7446615068452804e-07,
|
| 585 |
+
"logits/chosen": -1.6123485565185547,
|
| 586 |
+
"logits/rejected": -1.4534804821014404,
|
| 587 |
+
"logps/chosen": -510.5772399902344,
|
| 588 |
+
"logps/rejected": -681.8480224609375,
|
| 589 |
+
"loss": 0.4326,
|
| 590 |
"rewards/accuracies": 0.831250011920929,
|
| 591 |
+
"rewards/chosen": -2.398505687713623,
|
| 592 |
+
"rewards/margins": 1.5248647928237915,
|
| 593 |
+
"rewards/rejected": -3.923370838165283,
|
| 594 |
"step": 190
|
| 595 |
},
|
| 596 |
{
|
| 597 |
+
"epoch": 0.4116366514941619,
|
| 598 |
+
"grad_norm": 16.282324670439472,
|
| 599 |
"learning_rate": 3.6636955675673743e-07,
|
| 600 |
+
"logits/chosen": -1.6264712810516357,
|
| 601 |
+
"logits/rejected": -1.4750279188156128,
|
| 602 |
+
"logps/chosen": -514.038330078125,
|
| 603 |
+
"logps/rejected": -647.3781127929688,
|
| 604 |
+
"loss": 0.4404,
|
| 605 |
+
"rewards/accuracies": 0.8062499761581421,
|
| 606 |
+
"rewards/chosen": -2.37794828414917,
|
| 607 |
+
"rewards/margins": 1.2105674743652344,
|
| 608 |
+
"rewards/rejected": -3.5885162353515625,
|
| 609 |
"step": 195
|
| 610 |
},
|
| 611 |
{
|
| 612 |
+
"epoch": 0.4221914374299096,
|
| 613 |
+
"grad_norm": 21.312257244757074,
|
| 614 |
"learning_rate": 3.5811401601205093e-07,
|
| 615 |
+
"logits/chosen": -1.5191190242767334,
|
| 616 |
+
"logits/rejected": -1.5369209051132202,
|
| 617 |
+
"logps/chosen": -559.904052734375,
|
| 618 |
+
"logps/rejected": -721.0767211914062,
|
| 619 |
+
"loss": 0.4493,
|
| 620 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 621 |
+
"rewards/chosen": -2.823049545288086,
|
| 622 |
+
"rewards/margins": 1.5707520246505737,
|
| 623 |
+
"rewards/rejected": -4.393801689147949,
|
| 624 |
"step": 200
|
| 625 |
},
|
| 626 |
{
|
| 627 |
+
"epoch": 0.4327462233656574,
|
| 628 |
+
"grad_norm": 18.951174802021,
|
| 629 |
"learning_rate": 3.497108045260995e-07,
|
| 630 |
+
"logits/chosen": -1.6123138666152954,
|
| 631 |
+
"logits/rejected": -1.5209693908691406,
|
| 632 |
+
"logps/chosen": -508.78436279296875,
|
| 633 |
+
"logps/rejected": -680.7193603515625,
|
| 634 |
+
"loss": 0.4287,
|
| 635 |
+
"rewards/accuracies": 0.84375,
|
| 636 |
+
"rewards/chosen": -2.3978936672210693,
|
| 637 |
+
"rewards/margins": 1.566540002822876,
|
| 638 |
+
"rewards/rejected": -3.9644336700439453,
|
| 639 |
"step": 205
|
| 640 |
},
|
| 641 |
{
|
| 642 |
+
"epoch": 0.4433010093014051,
|
| 643 |
+
"grad_norm": 28.219925216993275,
|
| 644 |
"learning_rate": 3.411714000749838e-07,
|
| 645 |
+
"logits/chosen": -1.661116361618042,
|
| 646 |
+
"logits/rejected": -1.5592620372772217,
|
| 647 |
+
"logps/chosen": -547.822998046875,
|
| 648 |
+
"logps/rejected": -737.5689697265625,
|
| 649 |
+
"loss": 0.4248,
|
| 650 |
+
"rewards/accuracies": 0.84375,
|
| 651 |
+
"rewards/chosen": -2.826833724975586,
|
| 652 |
+
"rewards/margins": 1.727709412574768,
|
| 653 |
+
"rewards/rejected": -4.5545430183410645,
|
| 654 |
"step": 210
|
| 655 |
},
|
| 656 |
{
|
| 657 |
+
"epoch": 0.45385579523715286,
|
| 658 |
+
"grad_norm": 27.445717175281757,
|
| 659 |
"learning_rate": 3.3250746645801287e-07,
|
| 660 |
+
"logits/chosen": -1.6852436065673828,
|
| 661 |
+
"logits/rejected": -1.6252915859222412,
|
| 662 |
+
"logps/chosen": -603.220947265625,
|
| 663 |
+
"logps/rejected": -795.3065185546875,
|
| 664 |
+
"loss": 0.4409,
|
| 665 |
+
"rewards/accuracies": 0.800000011920929,
|
| 666 |
+
"rewards/chosen": -3.4767067432403564,
|
| 667 |
+
"rewards/margins": 1.7523845434188843,
|
| 668 |
+
"rewards/rejected": -5.229090690612793,
|
| 669 |
"step": 215
|
| 670 |
},
|
| 671 |
{
|
| 672 |
+
"epoch": 0.46441058117290057,
|
| 673 |
+
"grad_norm": 16.007752631407985,
|
| 674 |
"learning_rate": 3.237308375663571e-07,
|
| 675 |
+
"logits/chosen": -1.762291669845581,
|
| 676 |
+
"logits/rejected": -1.5295162200927734,
|
| 677 |
+
"logps/chosen": -531.2833862304688,
|
| 678 |
+
"logps/rejected": -725.7946166992188,
|
| 679 |
+
"loss": 0.3819,
|
| 680 |
+
"rewards/accuracies": 0.875,
|
| 681 |
+
"rewards/chosen": -2.7329602241516113,
|
| 682 |
+
"rewards/margins": 1.7636291980743408,
|
| 683 |
+
"rewards/rejected": -4.496589660644531,
|
| 684 |
"step": 220
|
| 685 |
},
|
| 686 |
{
|
| 687 |
+
"epoch": 0.47496536710864834,
|
| 688 |
+
"grad_norm": 18.506003391111292,
|
| 689 |
"learning_rate": 3.148535012193767e-07,
|
| 690 |
+
"logits/chosen": -1.754020094871521,
|
| 691 |
+
"logits/rejected": -1.6287786960601807,
|
| 692 |
+
"logps/chosen": -556.3197631835938,
|
| 693 |
+
"logps/rejected": -728.1083984375,
|
| 694 |
+
"loss": 0.3855,
|
| 695 |
+
"rewards/accuracies": 0.862500011920929,
|
| 696 |
+
"rewards/chosen": -2.739107370376587,
|
| 697 |
+
"rewards/margins": 1.6264499425888062,
|
| 698 |
+
"rewards/rejected": -4.3655571937561035,
|
| 699 |
"step": 225
|
| 700 |
},
|
| 701 |
{
|
| 702 |
+
"epoch": 0.48552015304439605,
|
| 703 |
+
"grad_norm": 20.68896480009483,
|
| 704 |
"learning_rate": 3.0588758279070183e-07,
|
| 705 |
+
"logits/chosen": -1.6177418231964111,
|
| 706 |
+
"logits/rejected": -1.5763094425201416,
|
| 707 |
+
"logps/chosen": -562.428955078125,
|
| 708 |
+
"logps/rejected": -753.3512573242188,
|
| 709 |
+
"loss": 0.4092,
|
| 710 |
"rewards/accuracies": 0.824999988079071,
|
| 711 |
+
"rewards/chosen": -2.9177470207214355,
|
| 712 |
+
"rewards/margins": 1.7720897197723389,
|
| 713 |
+
"rewards/rejected": -4.6898369789123535,
|
| 714 |
"step": 230
|
| 715 |
},
|
| 716 |
{
|
| 717 |
+
"epoch": 0.4960749389801438,
|
| 718 |
+
"grad_norm": 16.381736251716195,
|
| 719 |
"learning_rate": 2.968453286464312e-07,
|
| 720 |
+
"logits/chosen": -1.4933600425720215,
|
| 721 |
+
"logits/rejected": -1.5784003734588623,
|
| 722 |
+
"logps/chosen": -504.9471740722656,
|
| 723 |
+
"logps/rejected": -648.5574951171875,
|
| 724 |
+
"loss": 0.4179,
|
| 725 |
+
"rewards/accuracies": 0.762499988079071,
|
| 726 |
+
"rewards/chosen": -2.378976345062256,
|
| 727 |
+
"rewards/margins": 1.3860208988189697,
|
| 728 |
+
"rewards/rejected": -3.7649970054626465,
|
| 729 |
"step": 235
|
| 730 |
},
|
| 731 |
{
|
| 732 |
+
"epoch": 0.5066297249158915,
|
| 733 |
+
"grad_norm": 22.296205809275865,
|
| 734 |
"learning_rate": 2.8773908941806877e-07,
|
| 735 |
+
"logits/chosen": -1.6254231929779053,
|
| 736 |
+
"logits/rejected": -1.621469259262085,
|
| 737 |
+
"logps/chosen": -599.214599609375,
|
| 738 |
+
"logps/rejected": -809.9522705078125,
|
| 739 |
+
"loss": 0.39,
|
| 740 |
+
"rewards/accuracies": 0.84375,
|
| 741 |
+
"rewards/chosen": -3.3130722045898438,
|
| 742 |
+
"rewards/margins": 1.939295768737793,
|
| 743 |
+
"rewards/rejected": -5.2523674964904785,
|
| 744 |
"step": 240
|
| 745 |
},
|
| 746 |
{
|
| 747 |
+
"epoch": 0.5171845108516393,
|
| 748 |
+
"grad_norm": 17.974418608052964,
|
| 749 |
"learning_rate": 2.785813031330473e-07,
|
| 750 |
+
"logits/chosen": -1.6836649179458618,
|
| 751 |
+
"logits/rejected": -1.6946824789047241,
|
| 752 |
+
"logps/chosen": -636.5640869140625,
|
| 753 |
+
"logps/rejected": -862.1500244140625,
|
| 754 |
+
"loss": 0.407,
|
| 755 |
+
"rewards/accuracies": 0.8187500238418579,
|
| 756 |
+
"rewards/chosen": -3.709088087081909,
|
| 757 |
+
"rewards/margins": 2.079369068145752,
|
| 758 |
+
"rewards/rejected": -5.788456916809082,
|
| 759 |
"step": 245
|
| 760 |
},
|
| 761 |
{
|
| 762 |
+
"epoch": 0.5277392967873871,
|
| 763 |
+
"grad_norm": 21.869860477038394,
|
| 764 |
"learning_rate": 2.693844782258779e-07,
|
| 765 |
+
"logits/chosen": -1.597246766090393,
|
| 766 |
+
"logits/rejected": -1.4808999300003052,
|
| 767 |
+
"logps/chosen": -554.4249877929688,
|
| 768 |
+
"logps/rejected": -735.31103515625,
|
| 769 |
+
"loss": 0.377,
|
| 770 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 771 |
+
"rewards/chosen": -2.8226776123046875,
|
| 772 |
+
"rewards/margins": 1.673805832862854,
|
| 773 |
+
"rewards/rejected": -4.49648380279541,
|
| 774 |
"step": 250
|
| 775 |
},
|
| 776 |
{
|
| 777 |
+
"epoch": 0.5382940827231347,
|
| 778 |
+
"grad_norm": 35.342954704070486,
|
| 779 |
"learning_rate": 2.601611764531342e-07,
|
| 780 |
+
"logits/chosen": -1.599726676940918,
|
| 781 |
+
"logits/rejected": -1.5378262996673584,
|
| 782 |
+
"logps/chosen": -625.1596069335938,
|
| 783 |
+
"logps/rejected": -838.8173828125,
|
| 784 |
+
"loss": 0.3907,
|
| 785 |
+
"rewards/accuracies": 0.831250011920929,
|
| 786 |
+
"rewards/chosen": -3.5563435554504395,
|
| 787 |
+
"rewards/margins": 1.9609460830688477,
|
| 788 |
+
"rewards/rejected": -5.517289638519287,
|
| 789 |
"step": 255
|
| 790 |
},
|
| 791 |
{
|
| 792 |
+
"epoch": 0.5488488686588825,
|
| 793 |
+
"grad_norm": 17.9297074159325,
|
| 794 |
"learning_rate": 2.5092399573560323e-07,
|
| 795 |
+
"logits/chosen": -1.6411758661270142,
|
| 796 |
+
"logits/rejected": -1.6522302627563477,
|
| 797 |
+
"logps/chosen": -658.8689575195312,
|
| 798 |
+
"logps/rejected": -871.1637573242188,
|
| 799 |
+
"loss": 0.4168,
|
| 800 |
+
"rewards/accuracies": 0.8187500238418579,
|
| 801 |
+
"rewards/chosen": -3.866881847381592,
|
| 802 |
+
"rewards/margins": 2.113030195236206,
|
| 803 |
+
"rewards/rejected": -5.979912757873535,
|
| 804 |
"step": 260
|
| 805 |
},
|
| 806 |
{
|
| 807 |
+
"epoch": 0.5594036545946303,
|
| 808 |
+
"grad_norm": 16.443040045807887,
|
| 809 |
"learning_rate": 2.4168555295104124e-07,
|
| 810 |
+
"logits/chosen": -1.5852059125900269,
|
| 811 |
+
"logits/rejected": -1.5837304592132568,
|
| 812 |
+
"logps/chosen": -594.8711547851562,
|
| 813 |
+
"logps/rejected": -801.4606323242188,
|
| 814 |
+
"loss": 0.4093,
|
| 815 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 816 |
+
"rewards/chosen": -3.3196072578430176,
|
| 817 |
+
"rewards/margins": 1.8402855396270752,
|
| 818 |
+
"rewards/rejected": -5.159893035888672,
|
| 819 |
"step": 265
|
| 820 |
},
|
| 821 |
{
|
| 822 |
+
"epoch": 0.569958440530378,
|
| 823 |
+
"grad_norm": 15.035798283991243,
|
| 824 |
"learning_rate": 2.3245846670103626e-07,
|
| 825 |
+
"logits/chosen": -1.5612332820892334,
|
| 826 |
+
"logits/rejected": -1.5125606060028076,
|
| 827 |
+
"logps/chosen": -579.2890014648438,
|
| 828 |
+
"logps/rejected": -773.1954956054688,
|
| 829 |
+
"loss": 0.3769,
|
| 830 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 831 |
+
"rewards/chosen": -3.0682718753814697,
|
| 832 |
+
"rewards/margins": 1.7783126831054688,
|
| 833 |
+
"rewards/rejected": -4.846584320068359,
|
| 834 |
"step": 270
|
| 835 |
},
|
| 836 |
{
|
| 837 |
+
"epoch": 0.5805132264661257,
|
| 838 |
+
"grad_norm": 20.761167206822886,
|
| 839 |
"learning_rate": 2.232553400755159e-07,
|
| 840 |
+
"logits/chosen": -1.6236953735351562,
|
| 841 |
+
"logits/rejected": -1.5238358974456787,
|
| 842 |
+
"logps/chosen": -613.8088989257812,
|
| 843 |
+
"logps/rejected": -867.7283325195312,
|
| 844 |
+
"loss": 0.3697,
|
| 845 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 846 |
+
"rewards/chosen": -3.327275037765503,
|
| 847 |
+
"rewards/margins": 2.372307538986206,
|
| 848 |
+
"rewards/rejected": -5.699582576751709,
|
| 849 |
"step": 275
|
| 850 |
},
|
| 851 |
{
|
| 852 |
+
"epoch": 0.5910680124018735,
|
| 853 |
+
"grad_norm": 20.416122820074975,
|
| 854 |
"learning_rate": 2.1408874343844294e-07,
|
| 855 |
+
"logits/chosen": -1.6903560161590576,
|
| 856 |
+
"logits/rejected": -1.5515328645706177,
|
| 857 |
+
"logps/chosen": -648.6663818359375,
|
| 858 |
+
"logps/rejected": -949.74609375,
|
| 859 |
+
"loss": 0.3701,
|
| 860 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 861 |
+
"rewards/chosen": -3.67596435546875,
|
| 862 |
+
"rewards/margins": 2.625474214553833,
|
| 863 |
+
"rewards/rejected": -6.301438808441162,
|
| 864 |
"step": 280
|
| 865 |
},
|
| 866 |
{
|
| 867 |
+
"epoch": 0.6016227983376212,
|
| 868 |
+
"grad_norm": 20.954462877817495,
|
| 869 |
"learning_rate": 2.049711972582101e-07,
|
| 870 |
+
"logits/chosen": -1.7400896549224854,
|
| 871 |
+
"logits/rejected": -1.6318342685699463,
|
| 872 |
+
"logps/chosen": -647.7393798828125,
|
| 873 |
+
"logps/rejected": -899.9304809570312,
|
| 874 |
+
"loss": 0.3719,
|
| 875 |
+
"rewards/accuracies": 0.8125,
|
| 876 |
+
"rewards/chosen": -3.6654484272003174,
|
| 877 |
+
"rewards/margins": 2.371953010559082,
|
| 878 |
+
"rewards/rejected": -6.03740119934082,
|
| 879 |
"step": 285
|
| 880 |
},
|
| 881 |
{
|
| 882 |
+
"epoch": 0.612177584273369,
|
| 883 |
+
"grad_norm": 17.14928736018049,
|
| 884 |
"learning_rate": 1.9591515500618588e-07,
|
| 885 |
+
"logits/chosen": -1.6016016006469727,
|
| 886 |
+
"logits/rejected": -1.5183677673339844,
|
| 887 |
+
"logps/chosen": -670.8876342773438,
|
| 888 |
+
"logps/rejected": -865.2810668945312,
|
| 889 |
+
"loss": 0.4464,
|
| 890 |
+
"rewards/accuracies": 0.793749988079071,
|
| 891 |
+
"rewards/chosen": -3.901280641555786,
|
| 892 |
+
"rewards/margins": 1.737006425857544,
|
| 893 |
+
"rewards/rejected": -5.638287544250488,
|
| 894 |
"step": 290
|
| 895 |
},
|
| 896 |
{
|
| 897 |
+
"epoch": 0.6227323702091166,
|
| 898 |
+
"grad_norm": 15.631817925073218,
|
| 899 |
"learning_rate": 1.8693298614677112e-07,
|
| 900 |
+
"logits/chosen": -1.4730761051177979,
|
| 901 |
+
"logits/rejected": -1.3837854862213135,
|
| 902 |
+
"logps/chosen": -539.0037231445312,
|
| 903 |
+
"logps/rejected": -730.4061279296875,
|
| 904 |
+
"loss": 0.3841,
|
| 905 |
"rewards/accuracies": 0.875,
|
| 906 |
+
"rewards/chosen": -2.58305025100708,
|
| 907 |
+
"rewards/margins": 1.7921463251113892,
|
| 908 |
+
"rewards/rejected": -4.37519645690918,
|
| 909 |
"step": 295
|
| 910 |
},
|
| 911 |
{
|
| 912 |
+
"epoch": 0.6332871561448644,
|
| 913 |
+
"grad_norm": 17.780553626895177,
|
| 914 |
"learning_rate": 1.7803695924219814e-07,
|
| 915 |
+
"logits/chosen": -1.4669979810714722,
|
| 916 |
+
"logits/rejected": -1.4254872798919678,
|
| 917 |
+
"logps/chosen": -584.629150390625,
|
| 918 |
+
"logps/rejected": -784.3005981445312,
|
| 919 |
+
"loss": 0.4051,
|
| 920 |
+
"rewards/accuracies": 0.831250011920929,
|
| 921 |
+
"rewards/chosen": -3.1204633712768555,
|
| 922 |
+
"rewards/margins": 1.9275195598602295,
|
| 923 |
+
"rewards/rejected": -5.047983169555664,
|
| 924 |
"step": 300
|
| 925 |
},
|
| 926 |
{
|
| 927 |
+
"epoch": 0.6438419420806122,
|
| 928 |
+
"grad_norm": 19.866976638480853,
|
| 929 |
"learning_rate": 1.6923922519515067e-07,
|
| 930 |
+
"logits/chosen": -1.4443576335906982,
|
| 931 |
+
"logits/rejected": -1.4355580806732178,
|
| 932 |
+
"logps/chosen": -523.0612182617188,
|
| 933 |
+
"logps/rejected": -729.1703491210938,
|
| 934 |
+
"loss": 0.4069,
|
| 935 |
+
"rewards/accuracies": 0.84375,
|
| 936 |
+
"rewards/chosen": -2.749462842941284,
|
| 937 |
+
"rewards/margins": 1.8047136068344116,
|
| 938 |
+
"rewards/rejected": -4.554176330566406,
|
| 939 |
"step": 305
|
| 940 |
},
|
| 941 |
{
|
| 942 |
+
"epoch": 0.65439672801636,
|
| 943 |
+
"grad_norm": 19.42045257459303,
|
| 944 |
"learning_rate": 1.605518006520924e-07,
|
| 945 |
+
"logits/chosen": -1.4894784688949585,
|
| 946 |
+
"logits/rejected": -1.4559067487716675,
|
| 947 |
+
"logps/chosen": -589.7485961914062,
|
| 948 |
+
"logps/rejected": -808.5824584960938,
|
| 949 |
+
"loss": 0.3816,
|
| 950 |
+
"rewards/accuracies": 0.824999988079071,
|
| 951 |
+
"rewards/chosen": -3.1640734672546387,
|
| 952 |
+
"rewards/margins": 2.0186519622802734,
|
| 953 |
+
"rewards/rejected": -5.182725429534912,
|
| 954 |
"step": 310
|
| 955 |
},
|
| 956 |
{
|
| 957 |
+
"epoch": 0.6649515139521076,
|
| 958 |
+
"grad_norm": 34.470860837004935,
|
| 959 |
"learning_rate": 1.519865515899731e-07,
|
| 960 |
+
"logits/chosen": -1.5242574214935303,
|
| 961 |
+
"logits/rejected": -1.3928359746932983,
|
| 962 |
+
"logps/chosen": -588.0692138671875,
|
| 963 |
+
"logps/rejected": -841.7396240234375,
|
| 964 |
+
"loss": 0.3858,
|
| 965 |
+
"rewards/accuracies": 0.8125,
|
| 966 |
+
"rewards/chosen": -3.3201167583465576,
|
| 967 |
+
"rewards/margins": 2.1101181507110596,
|
| 968 |
+
"rewards/rejected": -5.430234432220459,
|
| 969 |
"step": 315
|
| 970 |
},
|
| 971 |
{
|
| 972 |
+
"epoch": 0.6755062998878554,
|
| 973 |
+
"grad_norm": 30.007742834102533,
|
| 974 |
"learning_rate": 1.4355517710873182e-07,
|
| 975 |
+
"logits/chosen": -1.546661615371704,
|
| 976 |
+
"logits/rejected": -1.4665791988372803,
|
| 977 |
+
"logps/chosen": -627.7368774414062,
|
| 978 |
+
"logps/rejected": -869.39453125,
|
| 979 |
+
"loss": 0.3914,
|
| 980 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 981 |
+
"rewards/chosen": -3.6285042762756348,
|
| 982 |
+
"rewards/margins": 2.2051749229431152,
|
| 983 |
+
"rewards/rejected": -5.83367919921875,
|
| 984 |
"step": 320
|
| 985 |
},
|
| 986 |
{
|
| 987 |
+
"epoch": 0.6860610858236031,
|
| 988 |
+
"grad_norm": 34.25630759725382,
|
| 989 |
"learning_rate": 1.3526919345173318e-07,
|
| 990 |
+
"logits/chosen": -1.5128138065338135,
|
| 991 |
+
"logits/rejected": -1.4411523342132568,
|
| 992 |
+
"logps/chosen": -602.6571044921875,
|
| 993 |
+
"logps/rejected": -849.2667846679688,
|
| 994 |
+
"loss": 0.4211,
|
| 995 |
"rewards/accuracies": 0.84375,
|
| 996 |
+
"rewards/chosen": -3.323326826095581,
|
| 997 |
+
"rewards/margins": 2.333648681640625,
|
| 998 |
+
"rewards/rejected": -5.656975269317627,
|
| 999 |
"step": 325
|
| 1000 |
},
|
| 1001 |
{
|
| 1002 |
+
"epoch": 0.6966158717593509,
|
| 1003 |
+
"grad_norm": 28.233731363069086,
|
| 1004 |
"learning_rate": 1.2713991827596443e-07,
|
| 1005 |
+
"logits/chosen": -1.495444655418396,
|
| 1006 |
+
"logits/rejected": -1.4902544021606445,
|
| 1007 |
+
"logps/chosen": -605.8360595703125,
|
| 1008 |
+
"logps/rejected": -796.5411376953125,
|
| 1009 |
+
"loss": 0.3791,
|
| 1010 |
+
"rewards/accuracies": 0.8187500238418579,
|
| 1011 |
+
"rewards/chosen": -3.165205478668213,
|
| 1012 |
+
"rewards/margins": 1.914367914199829,
|
| 1013 |
+
"rewards/rejected": -5.079573631286621,
|
| 1014 |
"step": 330
|
| 1015 |
},
|
| 1016 |
{
|
| 1017 |
+
"epoch": 0.7071706576950986,
|
| 1018 |
+
"grad_norm": 28.532009996830972,
|
| 1019 |
"learning_rate": 1.191784551934773e-07,
|
| 1020 |
+
"logits/chosen": -1.5311321020126343,
|
| 1021 |
+
"logits/rejected": -1.444746971130371,
|
| 1022 |
+
"logps/chosen": -515.9117431640625,
|
| 1023 |
+
"logps/rejected": -714.9118041992188,
|
| 1024 |
+
"loss": 0.3985,
|
| 1025 |
+
"rewards/accuracies": 0.862500011920929,
|
| 1026 |
+
"rewards/chosen": -2.741084575653076,
|
| 1027 |
+
"rewards/margins": 1.8353731632232666,
|
| 1028 |
+
"rewards/rejected": -4.5764570236206055,
|
| 1029 |
"step": 335
|
| 1030 |
},
|
| 1031 |
{
|
| 1032 |
+
"epoch": 0.7177254436308463,
|
| 1033 |
+
"grad_norm": 34.722496076575624,
|
| 1034 |
"learning_rate": 1.1139567860518953e-07,
|
| 1035 |
+
"logits/chosen": -1.3724250793457031,
|
| 1036 |
+
"logits/rejected": -1.3445093631744385,
|
| 1037 |
+
"logps/chosen": -516.8666381835938,
|
| 1038 |
+
"logps/rejected": -690.0352172851562,
|
| 1039 |
+
"loss": 0.4182,
|
| 1040 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1041 |
+
"rewards/chosen": -2.5948445796966553,
|
| 1042 |
+
"rewards/margins": 1.631908655166626,
|
| 1043 |
+
"rewards/rejected": -4.226753234863281,
|
| 1044 |
"step": 340
|
| 1045 |
},
|
| 1046 |
{
|
| 1047 |
+
"epoch": 0.7282802295665941,
|
| 1048 |
+
"grad_norm": 29.539595863207374,
|
| 1049 |
"learning_rate": 1.0380221884776128e-07,
|
| 1050 |
+
"logits/chosen": -1.4173920154571533,
|
| 1051 |
+
"logits/rejected": -1.3776549100875854,
|
| 1052 |
+
"logps/chosen": -542.7964477539062,
|
| 1053 |
+
"logps/rejected": -722.7305297851562,
|
| 1054 |
+
"loss": 0.4204,
|
| 1055 |
+
"rewards/accuracies": 0.856249988079071,
|
| 1056 |
+
"rewards/chosen": -2.7417469024658203,
|
| 1057 |
+
"rewards/margins": 1.6534277200698853,
|
| 1058 |
+
"rewards/rejected": -4.395174980163574,
|
| 1059 |
"step": 345
|
| 1060 |
},
|
| 1061 |
{
|
| 1062 |
+
"epoch": 0.7388350155023419,
|
| 1063 |
+
"grad_norm": 17.47998143760053,
|
| 1064 |
"learning_rate": 9.640844767383405e-08,
|
| 1065 |
+
"logits/chosen": -1.3427600860595703,
|
| 1066 |
+
"logits/rejected": -1.2479599714279175,
|
| 1067 |
+
"logps/chosen": -540.9366455078125,
|
| 1068 |
+
"logps/rejected": -741.2656860351562,
|
| 1069 |
+
"loss": 0.4206,
|
| 1070 |
+
"rewards/accuracies": 0.831250011920929,
|
| 1071 |
+
"rewards/chosen": -2.7238881587982178,
|
| 1072 |
+
"rewards/margins": 1.8479011058807373,
|
| 1073 |
+
"rewards/rejected": -4.571789741516113,
|
| 1074 |
"step": 350
|
| 1075 |
},
|
| 1076 |
{
|
| 1077 |
+
"epoch": 0.7493898014380895,
|
| 1078 |
+
"grad_norm": 22.53665974937468,
|
| 1079 |
"learning_rate": 8.922446408546378e-08,
|
| 1080 |
+
"logits/chosen": -1.3614610433578491,
|
| 1081 |
+
"logits/rejected": -1.3304545879364014,
|
| 1082 |
+
"logps/chosen": -564.0736694335938,
|
| 1083 |
+
"logps/rejected": -796.0197143554688,
|
| 1084 |
+
"loss": 0.4234,
|
| 1085 |
+
"rewards/accuracies": 0.8125,
|
| 1086 |
+
"rewards/chosen": -2.8989734649658203,
|
| 1087 |
+
"rewards/margins": 2.1152186393737793,
|
| 1088 |
+
"rewards/rejected": -5.0141921043396,
|
| 1089 |
"step": 355
|
| 1090 |
},
|
| 1091 |
{
|
| 1092 |
+
"epoch": 0.7599445873738373,
|
| 1093 |
+
"grad_norm": 34.016623930856696,
|
| 1094 |
"learning_rate": 8.22600805400994e-08,
|
| 1095 |
+
"logits/chosen": -1.4026951789855957,
|
| 1096 |
+
"logits/rejected": -1.3086416721343994,
|
| 1097 |
+
"logps/chosen": -557.1563720703125,
|
| 1098 |
+
"logps/rejected": -775.2821044921875,
|
| 1099 |
+
"loss": 0.3918,
|
| 1100 |
+
"rewards/accuracies": 0.8687499761581421,
|
| 1101 |
+
"rewards/chosen": -2.755949020385742,
|
| 1102 |
+
"rewards/margins": 1.9373924732208252,
|
| 1103 |
+
"rewards/rejected": -4.6933417320251465,
|
| 1104 |
"step": 360
|
| 1105 |
},
|
| 1106 |
{
|
| 1107 |
+
"epoch": 0.7704993733095851,
|
| 1108 |
+
"grad_norm": 19.583973928513963,
|
| 1109 |
"learning_rate": 7.552480954794558e-08,
|
| 1110 |
+
"logits/chosen": -1.4313266277313232,
|
| 1111 |
+
"logits/rejected": -1.3103513717651367,
|
| 1112 |
+
"logps/chosen": -564.1690673828125,
|
| 1113 |
+
"logps/rejected": -756.8692626953125,
|
| 1114 |
+
"loss": 0.4048,
|
| 1115 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 1116 |
+
"rewards/chosen": -2.88138484954834,
|
| 1117 |
+
"rewards/margins": 1.8533366918563843,
|
| 1118 |
+
"rewards/rejected": -4.7347211837768555,
|
| 1119 |
"step": 365
|
| 1120 |
},
|
| 1121 |
{
|
| 1122 |
+
"epoch": 0.7810541592453328,
|
| 1123 |
+
"grad_norm": 18.206536925706505,
|
| 1124 |
"learning_rate": 6.902785067901854e-08,
|
| 1125 |
+
"logits/chosen": -1.3645613193511963,
|
| 1126 |
+
"logits/rejected": -1.3341350555419922,
|
| 1127 |
+
"logps/chosen": -579.579833984375,
|
| 1128 |
+
"logps/rejected": -779.4511108398438,
|
| 1129 |
+
"loss": 0.3792,
|
| 1130 |
+
"rewards/accuracies": 0.8187500238418579,
|
| 1131 |
+
"rewards/chosen": -3.0597000122070312,
|
| 1132 |
+
"rewards/margins": 1.8546708822250366,
|
| 1133 |
+
"rewards/rejected": -4.914370536804199,
|
| 1134 |
"step": 370
|
| 1135 |
},
|
| 1136 |
{
|
| 1137 |
+
"epoch": 0.7916089451810805,
|
| 1138 |
+
"grad_norm": 25.99340540536939,
|
| 1139 |
"learning_rate": 6.277807799763973e-08,
|
| 1140 |
+
"logits/chosen": -1.5334607362747192,
|
| 1141 |
+
"logits/rejected": -1.4499049186706543,
|
| 1142 |
+
"logps/chosen": -563.6980590820312,
|
| 1143 |
+
"logps/rejected": -789.72119140625,
|
| 1144 |
+
"loss": 0.3903,
|
| 1145 |
+
"rewards/accuracies": 0.8687499761581421,
|
| 1146 |
+
"rewards/chosen": -3.1820693016052246,
|
| 1147 |
+
"rewards/margins": 1.9633289575576782,
|
| 1148 |
+
"rewards/rejected": -5.145398139953613,
|
| 1149 |
"step": 375
|
| 1150 |
},
|
| 1151 |
{
|
| 1152 |
+
"epoch": 0.8021637311168283,
|
| 1153 |
+
"grad_norm": 27.363349380647865,
|
| 1154 |
"learning_rate": 5.678402794153145e-08,
|
| 1155 |
+
"logits/chosen": -1.490678071975708,
|
| 1156 |
+
"logits/rejected": -1.4118311405181885,
|
| 1157 |
+
"logps/chosen": -635.3906860351562,
|
| 1158 |
+
"logps/rejected": -841.986328125,
|
| 1159 |
+
"loss": 0.3978,
|
| 1160 |
+
"rewards/accuracies": 0.824999988079071,
|
| 1161 |
+
"rewards/chosen": -3.6473312377929688,
|
| 1162 |
+
"rewards/margins": 1.8795642852783203,
|
| 1163 |
+
"rewards/rejected": -5.526895523071289,
|
| 1164 |
"step": 380
|
| 1165 |
},
|
| 1166 |
{
|
| 1167 |
+
"epoch": 0.812718517052576,
|
| 1168 |
+
"grad_norm": 22.024184023637442,
|
| 1169 |
"learning_rate": 5.105388766206969e-08,
|
| 1170 |
+
"logits/chosen": -1.6372134685516357,
|
| 1171 |
+
"logits/rejected": -1.4456651210784912,
|
| 1172 |
+
"logps/chosen": -670.1497802734375,
|
| 1173 |
+
"logps/rejected": -900.1561279296875,
|
| 1174 |
+
"loss": 0.4107,
|
| 1175 |
+
"rewards/accuracies": 0.84375,
|
| 1176 |
+
"rewards/chosen": -3.8800835609436035,
|
| 1177 |
+
"rewards/margins": 2.1356186866760254,
|
| 1178 |
+
"rewards/rejected": -6.015702724456787,
|
| 1179 |
"step": 385
|
| 1180 |
},
|
| 1181 |
{
|
| 1182 |
+
"epoch": 0.8232733029883238,
|
| 1183 |
+
"grad_norm": 18.957756630133876,
|
| 1184 |
"learning_rate": 4.5595483841620484e-08,
|
| 1185 |
+
"logits/chosen": -1.4949450492858887,
|
| 1186 |
+
"logits/rejected": -1.466933250427246,
|
| 1187 |
+
"logps/chosen": -592.1412963867188,
|
| 1188 |
+
"logps/rejected": -814.1785888671875,
|
| 1189 |
+
"loss": 0.3793,
|
| 1190 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 1191 |
+
"rewards/chosen": -3.3024775981903076,
|
| 1192 |
+
"rewards/margins": 1.9800523519515991,
|
| 1193 |
+
"rewards/rejected": -5.282530307769775,
|
| 1194 |
"step": 390
|
| 1195 |
},
|
| 1196 |
{
|
| 1197 |
+
"epoch": 0.8338280889240715,
|
| 1198 |
+
"grad_norm": 24.130033467348618,
|
| 1199 |
"learning_rate": 4.0416272003232526e-08,
|
| 1200 |
+
"logits/chosen": -1.5286778211593628,
|
| 1201 |
+
"logits/rejected": -1.4018694162368774,
|
| 1202 |
+
"logps/chosen": -614.2500610351562,
|
| 1203 |
+
"logps/rejected": -848.65673828125,
|
| 1204 |
+
"loss": 0.4233,
|
| 1205 |
"rewards/accuracies": 0.856249988079071,
|
| 1206 |
+
"rewards/chosen": -3.3793911933898926,
|
| 1207 |
+
"rewards/margins": 2.2501933574676514,
|
| 1208 |
+
"rewards/rejected": -5.629584312438965,
|
| 1209 |
"step": 395
|
| 1210 |
},
|
| 1211 |
{
|
| 1212 |
+
"epoch": 0.8443828748598192,
|
| 1213 |
+
"grad_norm": 21.924780448040742,
|
| 1214 |
"learning_rate": 3.552332632729041e-08,
|
| 1215 |
+
"logits/chosen": -1.3315622806549072,
|
| 1216 |
+
"logits/rejected": -1.4189374446868896,
|
| 1217 |
+
"logps/chosen": -591.8238525390625,
|
| 1218 |
+
"logps/rejected": -745.0551147460938,
|
| 1219 |
+
"loss": 0.4269,
|
| 1220 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1221 |
+
"rewards/chosen": -3.1886703968048096,
|
| 1222 |
+
"rewards/margins": 1.4928715229034424,
|
| 1223 |
+
"rewards/rejected": -4.68154239654541,
|
| 1224 |
"step": 400
|
| 1225 |
},
|
| 1226 |
{
|
| 1227 |
+
"epoch": 0.8443828748598192,
|
| 1228 |
+
"eval_logits/chosen": -1.3921160697937012,
|
| 1229 |
+
"eval_logits/rejected": -1.360321283340454,
|
| 1230 |
+
"eval_logps/chosen": -602.09814453125,
|
| 1231 |
+
"eval_logps/rejected": -797.6878051757812,
|
| 1232 |
+
"eval_loss": 0.38372838497161865,
|
| 1233 |
+
"eval_rewards/accuracies": 0.8643724918365479,
|
| 1234 |
+
"eval_rewards/chosen": -3.2510859966278076,
|
| 1235 |
+
"eval_rewards/margins": 1.8691294193267822,
|
| 1236 |
+
"eval_rewards/rejected": -5.120214939117432,
|
| 1237 |
+
"eval_runtime": 313.5439,
|
| 1238 |
+
"eval_samples_per_second": 6.302,
|
| 1239 |
+
"eval_steps_per_second": 1.576,
|
| 1240 |
"step": 400
|
| 1241 |
},
|
| 1242 |
{
|
| 1243 |
+
"epoch": 0.854937660795567,
|
| 1244 |
+
"grad_norm": 19.316578134763,
|
| 1245 |
"learning_rate": 3.092332998903416e-08,
|
| 1246 |
+
"logits/chosen": -1.4394410848617554,
|
| 1247 |
+
"logits/rejected": -1.3752410411834717,
|
| 1248 |
+
"logps/chosen": -601.6661987304688,
|
| 1249 |
+
"logps/rejected": -783.7593994140625,
|
| 1250 |
+
"loss": 0.3863,
|
| 1251 |
+
"rewards/accuracies": 0.8187500238418579,
|
| 1252 |
+
"rewards/chosen": -3.3254780769348145,
|
| 1253 |
+
"rewards/margins": 1.7059745788574219,
|
| 1254 |
+
"rewards/rejected": -5.031452655792236,
|
| 1255 |
"step": 405
|
| 1256 |
},
|
| 1257 |
{
|
| 1258 |
+
"epoch": 0.8654924467313148,
|
| 1259 |
+
"grad_norm": 28.757209433347523,
|
| 1260 |
"learning_rate": 2.6622566030146455e-08,
|
| 1261 |
+
"logits/chosen": -1.4320390224456787,
|
| 1262 |
+
"logits/rejected": -1.413570523262024,
|
| 1263 |
+
"logps/chosen": -565.2530517578125,
|
| 1264 |
+
"logps/rejected": -754.3922729492188,
|
| 1265 |
+
"loss": 0.4213,
|
| 1266 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 1267 |
+
"rewards/chosen": -2.987544298171997,
|
| 1268 |
+
"rewards/margins": 1.7588831186294556,
|
| 1269 |
+
"rewards/rejected": -4.7464280128479,
|
| 1270 |
"step": 410
|
| 1271 |
},
|
| 1272 |
{
|
| 1273 |
+
"epoch": 0.8760472326670625,
|
| 1274 |
+
"grad_norm": 18.43535305501894,
|
| 1275 |
"learning_rate": 2.26269087768734e-08,
|
| 1276 |
+
"logits/chosen": -1.4583203792572021,
|
| 1277 |
+
"logits/rejected": -1.3761074542999268,
|
| 1278 |
+
"logps/chosen": -577.720703125,
|
| 1279 |
+
"logps/rejected": -806.8515014648438,
|
| 1280 |
+
"loss": 0.3829,
|
| 1281 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 1282 |
+
"rewards/chosen": -3.2120673656463623,
|
| 1283 |
+
"rewards/margins": 2.084867477416992,
|
| 1284 |
+
"rewards/rejected": -5.296935081481934,
|
| 1285 |
"step": 415
|
| 1286 |
},
|
| 1287 |
{
|
| 1288 |
+
"epoch": 0.8866020186028102,
|
| 1289 |
+
"grad_norm": 20.96371705639774,
|
| 1290 |
"learning_rate": 1.894181581640106e-08,
|
| 1291 |
+
"logits/chosen": -1.384445071220398,
|
| 1292 |
+
"logits/rejected": -1.4241831302642822,
|
| 1293 |
+
"logps/chosen": -569.5962524414062,
|
| 1294 |
+
"logps/rejected": -749.0470581054688,
|
| 1295 |
+
"loss": 0.3819,
|
| 1296 |
+
"rewards/accuracies": 0.856249988079071,
|
| 1297 |
+
"rewards/chosen": -3.081413745880127,
|
| 1298 |
+
"rewards/margins": 1.7126514911651611,
|
| 1299 |
+
"rewards/rejected": -4.794064998626709,
|
| 1300 |
"step": 420
|
| 1301 |
},
|
| 1302 |
{
|
| 1303 |
+
"epoch": 0.897156804538558,
|
| 1304 |
+
"grad_norm": 26.21197116684321,
|
| 1305 |
"learning_rate": 1.5572320542448143e-08,
|
| 1306 |
+
"logits/chosen": -1.3703296184539795,
|
| 1307 |
+
"logits/rejected": -1.3327410221099854,
|
| 1308 |
+
"logps/chosen": -617.3011474609375,
|
| 1309 |
+
"logps/rejected": -836.2391357421875,
|
| 1310 |
+
"loss": 0.3825,
|
| 1311 |
+
"rewards/accuracies": 0.875,
|
| 1312 |
+
"rewards/chosen": -3.3691182136535645,
|
| 1313 |
+
"rewards/margins": 2.073474407196045,
|
| 1314 |
+
"rewards/rejected": -5.442592620849609,
|
| 1315 |
"step": 425
|
| 1316 |
},
|
| 1317 |
{
|
| 1318 |
+
"epoch": 0.9077115904743057,
|
| 1319 |
+
"grad_norm": 22.75331582162892,
|
| 1320 |
"learning_rate": 1.2523025280255729e-08,
|
| 1321 |
+
"logits/chosen": -1.4123306274414062,
|
| 1322 |
+
"logits/rejected": -1.4244548082351685,
|
| 1323 |
+
"logps/chosen": -593.9744262695312,
|
| 1324 |
+
"logps/rejected": -802.8662719726562,
|
| 1325 |
+
"loss": 0.3422,
|
| 1326 |
+
"rewards/accuracies": 0.8687499761581421,
|
| 1327 |
+
"rewards/chosen": -3.253312349319458,
|
| 1328 |
+
"rewards/margins": 1.9519180059432983,
|
| 1329 |
+
"rewards/rejected": -5.205230712890625,
|
| 1330 |
"step": 430
|
| 1331 |
},
|
| 1332 |
{
|
| 1333 |
+
"epoch": 0.9182663764100535,
|
| 1334 |
+
"grad_norm": 17.924586145441417,
|
| 1335 |
"learning_rate": 9.798095000364214e-09,
|
| 1336 |
+
"logits/chosen": -1.5641670227050781,
|
| 1337 |
+
"logits/rejected": -1.4084519147872925,
|
| 1338 |
+
"logps/chosen": -600.7659912109375,
|
| 1339 |
+
"logps/rejected": -869.5779418945312,
|
| 1340 |
+
"loss": 0.3553,
|
| 1341 |
+
"rewards/accuracies": 0.8812500238418579,
|
| 1342 |
+
"rewards/chosen": -3.2969956398010254,
|
| 1343 |
+
"rewards/margins": 2.371945381164551,
|
| 1344 |
+
"rewards/rejected": -5.668941020965576,
|
| 1345 |
"step": 435
|
| 1346 |
},
|
| 1347 |
{
|
| 1348 |
+
"epoch": 0.9288211623458011,
|
| 1349 |
+
"grad_norm": 25.268001512531786,
|
| 1350 |
"learning_rate": 7.401251629764876e-09,
|
| 1351 |
+
"logits/chosen": -1.5503554344177246,
|
| 1352 |
+
"logits/rejected": -1.3530725240707397,
|
| 1353 |
+
"logps/chosen": -560.983154296875,
|
| 1354 |
+
"logps/rejected": -772.5062255859375,
|
| 1355 |
+
"loss": 0.4149,
|
| 1356 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 1357 |
+
"rewards/chosen": -3.1528074741363525,
|
| 1358 |
+
"rewards/margins": 1.8247524499893188,
|
| 1359 |
+
"rewards/rejected": -4.977559566497803,
|
| 1360 |
"step": 440
|
| 1361 |
},
|
| 1362 |
{
|
| 1363 |
+
"epoch": 0.9393759482815489,
|
| 1364 |
+
"grad_norm": 18.848350951492684,
|
| 1365 |
"learning_rate": 5.335768968195098e-09,
|
| 1366 |
+
"logits/chosen": -1.4744082689285278,
|
| 1367 |
+
"logits/rejected": -1.370416283607483,
|
| 1368 |
+
"logps/chosen": -589.2518920898438,
|
| 1369 |
+
"logps/rejected": -806.16943359375,
|
| 1370 |
+
"loss": 0.4069,
|
| 1371 |
+
"rewards/accuracies": 0.856249988079071,
|
| 1372 |
+
"rewards/chosen": -3.1618690490722656,
|
| 1373 |
+
"rewards/margins": 2.018306255340576,
|
| 1374 |
+
"rewards/rejected": -5.180174827575684,
|
| 1375 |
"step": 445
|
| 1376 |
},
|
| 1377 |
{
|
| 1378 |
+
"epoch": 0.9499307342172967,
|
| 1379 |
+
"grad_norm": 14.44527353219914,
|
| 1380 |
"learning_rate": 3.604468216521883e-09,
|
| 1381 |
+
"logits/chosen": -1.3630057573318481,
|
| 1382 |
+
"logits/rejected": -1.2414896488189697,
|
| 1383 |
+
"logps/chosen": -591.0068359375,
|
| 1384 |
+
"logps/rejected": -788.7886352539062,
|
| 1385 |
+
"loss": 0.371,
|
| 1386 |
+
"rewards/accuracies": 0.875,
|
| 1387 |
+
"rewards/chosen": -3.234240770339966,
|
| 1388 |
+
"rewards/margins": 1.8283990621566772,
|
| 1389 |
+
"rewards/rejected": -5.0626397132873535,
|
| 1390 |
"step": 450
|
| 1391 |
},
|
| 1392 |
{
|
| 1393 |
+
"epoch": 0.9604855201530444,
|
| 1394 |
+
"grad_norm": 17.894278742200918,
|
| 1395 |
"learning_rate": 2.2097141233206884e-09,
|
| 1396 |
+
"logits/chosen": -1.5407589673995972,
|
| 1397 |
+
"logits/rejected": -1.4432313442230225,
|
| 1398 |
+
"logps/chosen": -616.9603271484375,
|
| 1399 |
+
"logps/rejected": -858.2979736328125,
|
| 1400 |
+
"loss": 0.3764,
|
| 1401 |
+
"rewards/accuracies": 0.856249988079071,
|
| 1402 |
+
"rewards/chosen": -3.5265698432922363,
|
| 1403 |
+
"rewards/margins": 2.2084403038024902,
|
| 1404 |
+
"rewards/rejected": -5.735010623931885,
|
| 1405 |
"step": 455
|
| 1406 |
},
|
| 1407 |
{
|
| 1408 |
+
"epoch": 0.9710403060887921,
|
| 1409 |
+
"grad_norm": 18.751457793725724,
|
| 1410 |
"learning_rate": 1.1534117549133472e-09,
|
| 1411 |
+
"logits/chosen": -1.46907639503479,
|
| 1412 |
+
"logits/rejected": -1.3426740169525146,
|
| 1413 |
+
"logps/chosen": -629.7129516601562,
|
| 1414 |
+
"logps/rejected": -850.1477661132812,
|
| 1415 |
+
"loss": 0.3512,
|
| 1416 |
+
"rewards/accuracies": 0.856249988079071,
|
| 1417 |
+
"rewards/chosen": -3.58038330078125,
|
| 1418 |
+
"rewards/margins": 2.005390167236328,
|
| 1419 |
+
"rewards/rejected": -5.585773468017578,
|
| 1420 |
"step": 460
|
| 1421 |
},
|
| 1422 |
{
|
| 1423 |
+
"epoch": 0.9815950920245399,
|
| 1424 |
+
"grad_norm": 29.39806840137654,
|
| 1425 |
"learning_rate": 4.3700389327672173e-10,
|
| 1426 |
+
"logits/chosen": -1.469868779182434,
|
| 1427 |
+
"logits/rejected": -1.3952000141143799,
|
| 1428 |
+
"logps/chosen": -604.9054565429688,
|
| 1429 |
+
"logps/rejected": -832.4953002929688,
|
| 1430 |
+
"loss": 0.3768,
|
| 1431 |
+
"rewards/accuracies": 0.8687499761581421,
|
| 1432 |
+
"rewards/chosen": -3.387165069580078,
|
| 1433 |
+
"rewards/margins": 2.0868310928344727,
|
| 1434 |
+
"rewards/rejected": -5.473996162414551,
|
| 1435 |
"step": 465
|
| 1436 |
},
|
| 1437 |
{
|
| 1438 |
+
"epoch": 0.9921498779602876,
|
| 1439 |
+
"grad_norm": 22.359148428267968,
|
| 1440 |
"learning_rate": 6.146906537587982e-11,
|
| 1441 |
+
"logits/chosen": -1.4432518482208252,
|
| 1442 |
+
"logits/rejected": -1.4169013500213623,
|
| 1443 |
+
"logps/chosen": -600.5186157226562,
|
| 1444 |
+
"logps/rejected": -811.341796875,
|
| 1445 |
+
"loss": 0.3939,
|
| 1446 |
+
"rewards/accuracies": 0.8062499761581421,
|
| 1447 |
+
"rewards/chosen": -3.3816261291503906,
|
| 1448 |
+
"rewards/margins": 2.0300629138946533,
|
| 1449 |
+
"rewards/rejected": -5.411688804626465,
|
| 1450 |
"step": 470
|
| 1451 |
},
|
| 1452 |
{
|
| 1453 |
+
"epoch": 0.9984827495217362,
|
| 1454 |
"step": 473,
|
| 1455 |
"total_flos": 0.0,
|
| 1456 |
+
"train_loss": 0.4645486564767285,
|
| 1457 |
+
"train_runtime": 24049.7915,
|
| 1458 |
+
"train_samples_per_second": 2.521,
|
| 1459 |
+
"train_steps_per_second": 0.02
|
| 1460 |
}
|
| 1461 |
],
|
| 1462 |
"logging_steps": 5,
|
|
|
|
| 1477 |
}
|
| 1478 |
},
|
| 1479 |
"total_flos": 0.0,
|
| 1480 |
+
"train_batch_size": 1,
|
| 1481 |
"trial_name": null,
|
| 1482 |
"trial_params": null
|
| 1483 |
}
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 7544
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0d1448ab9c2dd3a69fc51142481a50485c35957226f08c9032026a3fb5687e76
|
| 3 |
size 7544
|