Training in progress, step 4000, checkpoint
Browse files- checkpoint-4000/config.json +39 -0
- checkpoint-4000/generation_config.json +12 -0
- checkpoint-4000/model.safetensors +3 -0
- checkpoint-4000/optimizer.pt +3 -0
- checkpoint-4000/rng_state.pth +3 -0
- checkpoint-4000/scheduler.pt +3 -0
- checkpoint-4000/trainer_state.json +2834 -0
- checkpoint-4000/training_args.bin +3 -0
checkpoint-4000/config.json
ADDED
|
@@ -0,0 +1,39 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"LlamaForCausalLM"
|
| 4 |
+
],
|
| 5 |
+
"attention_bias": false,
|
| 6 |
+
"attention_dropout": 0.0,
|
| 7 |
+
"bos_token_id": 128000,
|
| 8 |
+
"eos_token_id": [
|
| 9 |
+
128001,
|
| 10 |
+
128008,
|
| 11 |
+
128009
|
| 12 |
+
],
|
| 13 |
+
"head_dim": 64,
|
| 14 |
+
"hidden_act": "silu",
|
| 15 |
+
"hidden_size": 2048,
|
| 16 |
+
"initializer_range": 0.02,
|
| 17 |
+
"intermediate_size": 8192,
|
| 18 |
+
"max_position_embeddings": 131072,
|
| 19 |
+
"mlp_bias": false,
|
| 20 |
+
"model_type": "llama",
|
| 21 |
+
"num_attention_heads": 32,
|
| 22 |
+
"num_hidden_layers": 16,
|
| 23 |
+
"num_key_value_heads": 8,
|
| 24 |
+
"pretraining_tp": 1,
|
| 25 |
+
"rms_norm_eps": 1e-05,
|
| 26 |
+
"rope_scaling": {
|
| 27 |
+
"factor": 32.0,
|
| 28 |
+
"high_freq_factor": 4.0,
|
| 29 |
+
"low_freq_factor": 1.0,
|
| 30 |
+
"original_max_position_embeddings": 8192,
|
| 31 |
+
"rope_type": "llama3"
|
| 32 |
+
},
|
| 33 |
+
"rope_theta": 500000.0,
|
| 34 |
+
"tie_word_embeddings": true,
|
| 35 |
+
"torch_dtype": "bfloat16",
|
| 36 |
+
"transformers_version": "4.51.3",
|
| 37 |
+
"use_cache": true,
|
| 38 |
+
"vocab_size": 128256
|
| 39 |
+
}
|
checkpoint-4000/generation_config.json
ADDED
|
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token_id": 128000,
|
| 3 |
+
"do_sample": true,
|
| 4 |
+
"eos_token_id": [
|
| 5 |
+
128001,
|
| 6 |
+
128008,
|
| 7 |
+
128009
|
| 8 |
+
],
|
| 9 |
+
"temperature": 0.6,
|
| 10 |
+
"top_p": 0.9,
|
| 11 |
+
"transformers_version": "4.51.3"
|
| 12 |
+
}
|
checkpoint-4000/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1a97bb800e1aa2969f1dd45315d0b4fdba460f2e15ef636509c476b4091eabf1
|
| 3 |
+
size 2471645608
|
checkpoint-4000/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:73eb527a3024fcab9157164d6f789fc19bbd2bcb1464e23dd7b4c9ab6b008034
|
| 3 |
+
size 3724602
|
checkpoint-4000/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d1ab5d0a7b221ebbbfd24892671a396889e139663559496bbbfc652af6cd8113
|
| 3 |
+
size 14180
|
checkpoint-4000/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:731316ea7a5c258bcaac030cb3666cbae5e4a279bb34409b3f0dcbb7bbfc41b5
|
| 3 |
+
size 1064
|
checkpoint-4000/trainer_state.json
ADDED
|
@@ -0,0 +1,2834 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 1.0,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 4000,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.0025,
|
| 14 |
+
"grad_norm": 276.0,
|
| 15 |
+
"learning_rate": 1.1250000000000001e-07,
|
| 16 |
+
"loss": 0.4619,
|
| 17 |
+
"step": 10
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.005,
|
| 21 |
+
"grad_norm": 236.0,
|
| 22 |
+
"learning_rate": 2.3750000000000003e-07,
|
| 23 |
+
"loss": 0.457,
|
| 24 |
+
"step": 20
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.0075,
|
| 28 |
+
"grad_norm": 276.0,
|
| 29 |
+
"learning_rate": 3.625e-07,
|
| 30 |
+
"loss": 0.51,
|
| 31 |
+
"step": 30
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.01,
|
| 35 |
+
"grad_norm": 194.0,
|
| 36 |
+
"learning_rate": 4.875000000000001e-07,
|
| 37 |
+
"loss": 0.4631,
|
| 38 |
+
"step": 40
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.0125,
|
| 42 |
+
"grad_norm": 264.0,
|
| 43 |
+
"learning_rate": 6.125000000000001e-07,
|
| 44 |
+
"loss": 0.5266,
|
| 45 |
+
"step": 50
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.015,
|
| 49 |
+
"grad_norm": 21.75,
|
| 50 |
+
"learning_rate": 7.375e-07,
|
| 51 |
+
"loss": 0.5422,
|
| 52 |
+
"step": 60
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.0175,
|
| 56 |
+
"grad_norm": 21.875,
|
| 57 |
+
"learning_rate": 8.625e-07,
|
| 58 |
+
"loss": 0.5314,
|
| 59 |
+
"step": 70
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.02,
|
| 63 |
+
"grad_norm": 170.0,
|
| 64 |
+
"learning_rate": 9.875e-07,
|
| 65 |
+
"loss": 0.4754,
|
| 66 |
+
"step": 80
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.0225,
|
| 70 |
+
"grad_norm": 5.625,
|
| 71 |
+
"learning_rate": 1.1125000000000001e-06,
|
| 72 |
+
"loss": 0.5029,
|
| 73 |
+
"step": 90
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.025,
|
| 77 |
+
"grad_norm": 278.0,
|
| 78 |
+
"learning_rate": 1.2375e-06,
|
| 79 |
+
"loss": 0.4709,
|
| 80 |
+
"step": 100
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.0275,
|
| 84 |
+
"grad_norm": 2.546875,
|
| 85 |
+
"learning_rate": 1.3625000000000003e-06,
|
| 86 |
+
"loss": 0.3525,
|
| 87 |
+
"step": 110
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.03,
|
| 91 |
+
"grad_norm": 74.0,
|
| 92 |
+
"learning_rate": 1.4875000000000002e-06,
|
| 93 |
+
"loss": 0.3973,
|
| 94 |
+
"step": 120
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.0325,
|
| 98 |
+
"grad_norm": 2.4375,
|
| 99 |
+
"learning_rate": 1.6125000000000002e-06,
|
| 100 |
+
"loss": 0.3656,
|
| 101 |
+
"step": 130
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.035,
|
| 105 |
+
"grad_norm": 232.0,
|
| 106 |
+
"learning_rate": 1.7375e-06,
|
| 107 |
+
"loss": 0.3996,
|
| 108 |
+
"step": 140
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.0375,
|
| 112 |
+
"grad_norm": 4.09375,
|
| 113 |
+
"learning_rate": 1.8625000000000002e-06,
|
| 114 |
+
"loss": 0.3652,
|
| 115 |
+
"step": 150
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.04,
|
| 119 |
+
"grad_norm": 12.875,
|
| 120 |
+
"learning_rate": 1.9875000000000005e-06,
|
| 121 |
+
"loss": 0.4156,
|
| 122 |
+
"step": 160
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.0425,
|
| 126 |
+
"grad_norm": 22.625,
|
| 127 |
+
"learning_rate": 2.1125e-06,
|
| 128 |
+
"loss": 0.3318,
|
| 129 |
+
"step": 170
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.045,
|
| 133 |
+
"grad_norm": 12.75,
|
| 134 |
+
"learning_rate": 2.2375e-06,
|
| 135 |
+
"loss": 0.3256,
|
| 136 |
+
"step": 180
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.0475,
|
| 140 |
+
"grad_norm": 5.09375,
|
| 141 |
+
"learning_rate": 2.3625000000000003e-06,
|
| 142 |
+
"loss": 0.3281,
|
| 143 |
+
"step": 190
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.05,
|
| 147 |
+
"grad_norm": 25.875,
|
| 148 |
+
"learning_rate": 2.4875000000000003e-06,
|
| 149 |
+
"loss": 0.3357,
|
| 150 |
+
"step": 200
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.0525,
|
| 154 |
+
"grad_norm": 2.703125,
|
| 155 |
+
"learning_rate": 2.6125e-06,
|
| 156 |
+
"loss": 0.3479,
|
| 157 |
+
"step": 210
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 0.055,
|
| 161 |
+
"grad_norm": 2.765625,
|
| 162 |
+
"learning_rate": 2.7375e-06,
|
| 163 |
+
"loss": 0.3328,
|
| 164 |
+
"step": 220
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.0575,
|
| 168 |
+
"grad_norm": 2.0,
|
| 169 |
+
"learning_rate": 2.8625e-06,
|
| 170 |
+
"loss": 0.3164,
|
| 171 |
+
"step": 230
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.06,
|
| 175 |
+
"grad_norm": 2.34375,
|
| 176 |
+
"learning_rate": 2.9875e-06,
|
| 177 |
+
"loss": 0.299,
|
| 178 |
+
"step": 240
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.0625,
|
| 182 |
+
"grad_norm": 2.84375,
|
| 183 |
+
"learning_rate": 3.1125000000000007e-06,
|
| 184 |
+
"loss": 0.3129,
|
| 185 |
+
"step": 250
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.065,
|
| 189 |
+
"grad_norm": 2.671875,
|
| 190 |
+
"learning_rate": 3.2375e-06,
|
| 191 |
+
"loss": 0.3131,
|
| 192 |
+
"step": 260
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.0675,
|
| 196 |
+
"grad_norm": 4.09375,
|
| 197 |
+
"learning_rate": 3.3625000000000004e-06,
|
| 198 |
+
"loss": 0.3053,
|
| 199 |
+
"step": 270
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.07,
|
| 203 |
+
"grad_norm": 1.9375,
|
| 204 |
+
"learning_rate": 3.4875000000000005e-06,
|
| 205 |
+
"loss": 0.299,
|
| 206 |
+
"step": 280
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.0725,
|
| 210 |
+
"grad_norm": 2.6875,
|
| 211 |
+
"learning_rate": 3.6125000000000006e-06,
|
| 212 |
+
"loss": 0.3273,
|
| 213 |
+
"step": 290
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.075,
|
| 217 |
+
"grad_norm": 1.90625,
|
| 218 |
+
"learning_rate": 3.7375000000000006e-06,
|
| 219 |
+
"loss": 0.3041,
|
| 220 |
+
"step": 300
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.0775,
|
| 224 |
+
"grad_norm": 1.640625,
|
| 225 |
+
"learning_rate": 3.8625e-06,
|
| 226 |
+
"loss": 0.3057,
|
| 227 |
+
"step": 310
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.08,
|
| 231 |
+
"grad_norm": 2.375,
|
| 232 |
+
"learning_rate": 3.9875e-06,
|
| 233 |
+
"loss": 0.315,
|
| 234 |
+
"step": 320
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.0825,
|
| 238 |
+
"grad_norm": 2.015625,
|
| 239 |
+
"learning_rate": 4.1125e-06,
|
| 240 |
+
"loss": 0.3063,
|
| 241 |
+
"step": 330
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.085,
|
| 245 |
+
"grad_norm": 1.96875,
|
| 246 |
+
"learning_rate": 4.2375000000000005e-06,
|
| 247 |
+
"loss": 0.3023,
|
| 248 |
+
"step": 340
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.0875,
|
| 252 |
+
"grad_norm": 2.53125,
|
| 253 |
+
"learning_rate": 4.362500000000001e-06,
|
| 254 |
+
"loss": 0.2834,
|
| 255 |
+
"step": 350
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.09,
|
| 259 |
+
"grad_norm": 2.0,
|
| 260 |
+
"learning_rate": 4.4875e-06,
|
| 261 |
+
"loss": 0.2767,
|
| 262 |
+
"step": 360
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.0925,
|
| 266 |
+
"grad_norm": 1.6015625,
|
| 267 |
+
"learning_rate": 4.6125e-06,
|
| 268 |
+
"loss": 0.2859,
|
| 269 |
+
"step": 370
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.095,
|
| 273 |
+
"grad_norm": 2.375,
|
| 274 |
+
"learning_rate": 4.737500000000001e-06,
|
| 275 |
+
"loss": 0.2889,
|
| 276 |
+
"step": 380
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.0975,
|
| 280 |
+
"grad_norm": 4.75,
|
| 281 |
+
"learning_rate": 4.8625000000000005e-06,
|
| 282 |
+
"loss": 0.2916,
|
| 283 |
+
"step": 390
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.1,
|
| 287 |
+
"grad_norm": 1.671875,
|
| 288 |
+
"learning_rate": 4.987500000000001e-06,
|
| 289 |
+
"loss": 0.2722,
|
| 290 |
+
"step": 400
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.1025,
|
| 294 |
+
"grad_norm": 2.46875,
|
| 295 |
+
"learning_rate": 4.999922894111975e-06,
|
| 296 |
+
"loss": 0.2822,
|
| 297 |
+
"step": 410
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.105,
|
| 301 |
+
"grad_norm": 2.515625,
|
| 302 |
+
"learning_rate": 4.999656361346094e-06,
|
| 303 |
+
"loss": 0.269,
|
| 304 |
+
"step": 420
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 0.1075,
|
| 308 |
+
"grad_norm": 1.6875,
|
| 309 |
+
"learning_rate": 4.999199470070484e-06,
|
| 310 |
+
"loss": 0.2836,
|
| 311 |
+
"step": 430
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 0.11,
|
| 315 |
+
"grad_norm": 1.671875,
|
| 316 |
+
"learning_rate": 4.998552255079182e-06,
|
| 317 |
+
"loss": 0.2577,
|
| 318 |
+
"step": 440
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.1125,
|
| 322 |
+
"grad_norm": 1.5234375,
|
| 323 |
+
"learning_rate": 4.9977147656601196e-06,
|
| 324 |
+
"loss": 0.266,
|
| 325 |
+
"step": 450
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.115,
|
| 329 |
+
"grad_norm": 1.890625,
|
| 330 |
+
"learning_rate": 4.996687065591355e-06,
|
| 331 |
+
"loss": 0.2792,
|
| 332 |
+
"step": 460
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.1175,
|
| 336 |
+
"grad_norm": 2.484375,
|
| 337 |
+
"learning_rate": 4.9954692331362295e-06,
|
| 338 |
+
"loss": 0.2801,
|
| 339 |
+
"step": 470
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.12,
|
| 343 |
+
"grad_norm": 1.828125,
|
| 344 |
+
"learning_rate": 4.9940613610373974e-06,
|
| 345 |
+
"loss": 0.26,
|
| 346 |
+
"step": 480
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.1225,
|
| 350 |
+
"grad_norm": 1.9296875,
|
| 351 |
+
"learning_rate": 4.992463556509772e-06,
|
| 352 |
+
"loss": 0.2613,
|
| 353 |
+
"step": 490
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 0.125,
|
| 357 |
+
"grad_norm": 1.828125,
|
| 358 |
+
"learning_rate": 4.990675941232353e-06,
|
| 359 |
+
"loss": 0.2385,
|
| 360 |
+
"step": 500
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 0.1275,
|
| 364 |
+
"grad_norm": 1.90625,
|
| 365 |
+
"learning_rate": 4.988698651338965e-06,
|
| 366 |
+
"loss": 0.2574,
|
| 367 |
+
"step": 510
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 0.13,
|
| 371 |
+
"grad_norm": 1.5,
|
| 372 |
+
"learning_rate": 4.986531837407891e-06,
|
| 373 |
+
"loss": 0.2559,
|
| 374 |
+
"step": 520
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.1325,
|
| 378 |
+
"grad_norm": 1.484375,
|
| 379 |
+
"learning_rate": 4.9841756644503965e-06,
|
| 380 |
+
"loss": 0.2622,
|
| 381 |
+
"step": 530
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.135,
|
| 385 |
+
"grad_norm": 1.15625,
|
| 386 |
+
"learning_rate": 4.981630311898178e-06,
|
| 387 |
+
"loss": 0.2537,
|
| 388 |
+
"step": 540
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.1375,
|
| 392 |
+
"grad_norm": 1.6484375,
|
| 393 |
+
"learning_rate": 4.978895973589686e-06,
|
| 394 |
+
"loss": 0.2832,
|
| 395 |
+
"step": 550
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 0.14,
|
| 399 |
+
"grad_norm": 1.6796875,
|
| 400 |
+
"learning_rate": 4.975972857755369e-06,
|
| 401 |
+
"loss": 0.2471,
|
| 402 |
+
"step": 560
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 0.1425,
|
| 406 |
+
"grad_norm": 1.4296875,
|
| 407 |
+
"learning_rate": 4.972861187001815e-06,
|
| 408 |
+
"loss": 0.2648,
|
| 409 |
+
"step": 570
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 0.145,
|
| 413 |
+
"grad_norm": 1.6171875,
|
| 414 |
+
"learning_rate": 4.9695611982947995e-06,
|
| 415 |
+
"loss": 0.2697,
|
| 416 |
+
"step": 580
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.1475,
|
| 420 |
+
"grad_norm": 1.8515625,
|
| 421 |
+
"learning_rate": 4.966073142941239e-06,
|
| 422 |
+
"loss": 0.2592,
|
| 423 |
+
"step": 590
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.15,
|
| 427 |
+
"grad_norm": 1.4140625,
|
| 428 |
+
"learning_rate": 4.962397286570053e-06,
|
| 429 |
+
"loss": 0.267,
|
| 430 |
+
"step": 600
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 0.1525,
|
| 434 |
+
"grad_norm": 1.515625,
|
| 435 |
+
"learning_rate": 4.958533909111936e-06,
|
| 436 |
+
"loss": 0.253,
|
| 437 |
+
"step": 610
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 0.155,
|
| 441 |
+
"grad_norm": 1.390625,
|
| 442 |
+
"learning_rate": 4.95448330477804e-06,
|
| 443 |
+
"loss": 0.2488,
|
| 444 |
+
"step": 620
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.1575,
|
| 448 |
+
"grad_norm": 1.71875,
|
| 449 |
+
"learning_rate": 4.950245782037566e-06,
|
| 450 |
+
"loss": 0.2552,
|
| 451 |
+
"step": 630
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 0.16,
|
| 455 |
+
"grad_norm": 1.2890625,
|
| 456 |
+
"learning_rate": 4.945821663594277e-06,
|
| 457 |
+
"loss": 0.2485,
|
| 458 |
+
"step": 640
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 0.1625,
|
| 462 |
+
"grad_norm": 1.5078125,
|
| 463 |
+
"learning_rate": 4.941211286361922e-06,
|
| 464 |
+
"loss": 0.2612,
|
| 465 |
+
"step": 650
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 0.165,
|
| 469 |
+
"grad_norm": 1.8984375,
|
| 470 |
+
"learning_rate": 4.936415001438577e-06,
|
| 471 |
+
"loss": 0.2465,
|
| 472 |
+
"step": 660
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 0.1675,
|
| 476 |
+
"grad_norm": 1.609375,
|
| 477 |
+
"learning_rate": 4.9314331740799084e-06,
|
| 478 |
+
"loss": 0.2449,
|
| 479 |
+
"step": 670
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 0.17,
|
| 483 |
+
"grad_norm": 1.234375,
|
| 484 |
+
"learning_rate": 4.926266183671356e-06,
|
| 485 |
+
"loss": 0.2431,
|
| 486 |
+
"step": 680
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 0.1725,
|
| 490 |
+
"grad_norm": 1.359375,
|
| 491 |
+
"learning_rate": 4.920914423699247e-06,
|
| 492 |
+
"loss": 0.2479,
|
| 493 |
+
"step": 690
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 0.175,
|
| 497 |
+
"grad_norm": 1.28125,
|
| 498 |
+
"learning_rate": 4.915378301720822e-06,
|
| 499 |
+
"loss": 0.2327,
|
| 500 |
+
"step": 700
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 0.1775,
|
| 504 |
+
"grad_norm": 1.3984375,
|
| 505 |
+
"learning_rate": 4.909658239333203e-06,
|
| 506 |
+
"loss": 0.2434,
|
| 507 |
+
"step": 710
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 0.18,
|
| 511 |
+
"grad_norm": 1.4296875,
|
| 512 |
+
"learning_rate": 4.903754672141288e-06,
|
| 513 |
+
"loss": 0.2437,
|
| 514 |
+
"step": 720
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 0.1825,
|
| 518 |
+
"grad_norm": 1.6953125,
|
| 519 |
+
"learning_rate": 4.897668049724574e-06,
|
| 520 |
+
"loss": 0.248,
|
| 521 |
+
"step": 730
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 0.185,
|
| 525 |
+
"grad_norm": 1.3515625,
|
| 526 |
+
"learning_rate": 4.891398835602925e-06,
|
| 527 |
+
"loss": 0.2502,
|
| 528 |
+
"step": 740
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 0.1875,
|
| 532 |
+
"grad_norm": 1.5546875,
|
| 533 |
+
"learning_rate": 4.884947507201268e-06,
|
| 534 |
+
"loss": 0.2368,
|
| 535 |
+
"step": 750
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 0.19,
|
| 539 |
+
"grad_norm": 1.7421875,
|
| 540 |
+
"learning_rate": 4.878314555813237e-06,
|
| 541 |
+
"loss": 0.2262,
|
| 542 |
+
"step": 760
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 0.1925,
|
| 546 |
+
"grad_norm": 1.6328125,
|
| 547 |
+
"learning_rate": 4.8715004865637616e-06,
|
| 548 |
+
"loss": 0.2392,
|
| 549 |
+
"step": 770
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.195,
|
| 553 |
+
"grad_norm": 2.015625,
|
| 554 |
+
"learning_rate": 4.8645058183705976e-06,
|
| 555 |
+
"loss": 0.2403,
|
| 556 |
+
"step": 780
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 0.1975,
|
| 560 |
+
"grad_norm": 1.2109375,
|
| 561 |
+
"learning_rate": 4.8573310839048085e-06,
|
| 562 |
+
"loss": 0.2466,
|
| 563 |
+
"step": 790
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 0.2,
|
| 567 |
+
"grad_norm": 1.3125,
|
| 568 |
+
"learning_rate": 4.8499768295502e-06,
|
| 569 |
+
"loss": 0.2453,
|
| 570 |
+
"step": 800
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 0.2025,
|
| 574 |
+
"grad_norm": 2.03125,
|
| 575 |
+
"learning_rate": 4.842443615361718e-06,
|
| 576 |
+
"loss": 0.2566,
|
| 577 |
+
"step": 810
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 0.205,
|
| 581 |
+
"grad_norm": 1.9296875,
|
| 582 |
+
"learning_rate": 4.834732015022786e-06,
|
| 583 |
+
"loss": 0.2248,
|
| 584 |
+
"step": 820
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 0.2075,
|
| 588 |
+
"grad_norm": 1.59375,
|
| 589 |
+
"learning_rate": 4.826842615801628e-06,
|
| 590 |
+
"loss": 0.2456,
|
| 591 |
+
"step": 830
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 0.21,
|
| 595 |
+
"grad_norm": 1.484375,
|
| 596 |
+
"learning_rate": 4.818776018506538e-06,
|
| 597 |
+
"loss": 0.2445,
|
| 598 |
+
"step": 840
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 0.2125,
|
| 602 |
+
"grad_norm": 1.5625,
|
| 603 |
+
"learning_rate": 4.810532837440134e-06,
|
| 604 |
+
"loss": 0.2451,
|
| 605 |
+
"step": 850
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 0.215,
|
| 609 |
+
"grad_norm": 1.5390625,
|
| 610 |
+
"learning_rate": 4.802113700352567e-06,
|
| 611 |
+
"loss": 0.2605,
|
| 612 |
+
"step": 860
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 0.2175,
|
| 616 |
+
"grad_norm": 1.2734375,
|
| 617 |
+
"learning_rate": 4.793519248393721e-06,
|
| 618 |
+
"loss": 0.2394,
|
| 619 |
+
"step": 870
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 0.22,
|
| 623 |
+
"grad_norm": 1.203125,
|
| 624 |
+
"learning_rate": 4.78475013606439e-06,
|
| 625 |
+
"loss": 0.238,
|
| 626 |
+
"step": 880
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 0.2225,
|
| 630 |
+
"grad_norm": 1.3671875,
|
| 631 |
+
"learning_rate": 4.775807031166428e-06,
|
| 632 |
+
"loss": 0.236,
|
| 633 |
+
"step": 890
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 0.225,
|
| 637 |
+
"grad_norm": 1.7578125,
|
| 638 |
+
"learning_rate": 4.766690614751897e-06,
|
| 639 |
+
"loss": 0.2357,
|
| 640 |
+
"step": 900
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 0.2275,
|
| 644 |
+
"grad_norm": 1.3359375,
|
| 645 |
+
"learning_rate": 4.757401581071203e-06,
|
| 646 |
+
"loss": 0.2436,
|
| 647 |
+
"step": 910
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 0.23,
|
| 651 |
+
"grad_norm": 3.671875,
|
| 652 |
+
"learning_rate": 4.747940637520226e-06,
|
| 653 |
+
"loss": 0.2309,
|
| 654 |
+
"step": 920
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.2325,
|
| 658 |
+
"grad_norm": 1.40625,
|
| 659 |
+
"learning_rate": 4.738308504586445e-06,
|
| 660 |
+
"loss": 0.2254,
|
| 661 |
+
"step": 930
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 0.235,
|
| 665 |
+
"grad_norm": 1.359375,
|
| 666 |
+
"learning_rate": 4.7285059157940765e-06,
|
| 667 |
+
"loss": 0.2221,
|
| 668 |
+
"step": 940
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 0.2375,
|
| 672 |
+
"grad_norm": 1.1640625,
|
| 673 |
+
"learning_rate": 4.718533617648209e-06,
|
| 674 |
+
"loss": 0.2246,
|
| 675 |
+
"step": 950
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 0.24,
|
| 679 |
+
"grad_norm": 1.1171875,
|
| 680 |
+
"learning_rate": 4.7083923695779546e-06,
|
| 681 |
+
"loss": 0.2228,
|
| 682 |
+
"step": 960
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 0.2425,
|
| 686 |
+
"grad_norm": 1.7109375,
|
| 687 |
+
"learning_rate": 4.6980829438786176e-06,
|
| 688 |
+
"loss": 0.2198,
|
| 689 |
+
"step": 970
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 0.245,
|
| 693 |
+
"grad_norm": 1.9453125,
|
| 694 |
+
"learning_rate": 4.687606125652882e-06,
|
| 695 |
+
"loss": 0.223,
|
| 696 |
+
"step": 980
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 0.2475,
|
| 700 |
+
"grad_norm": 1.5,
|
| 701 |
+
"learning_rate": 4.676962712751015e-06,
|
| 702 |
+
"loss": 0.2337,
|
| 703 |
+
"step": 990
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 0.25,
|
| 707 |
+
"grad_norm": 1.359375,
|
| 708 |
+
"learning_rate": 4.666153515710118e-06,
|
| 709 |
+
"loss": 0.2273,
|
| 710 |
+
"step": 1000
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 0.2525,
|
| 714 |
+
"grad_norm": 1.2109375,
|
| 715 |
+
"learning_rate": 4.655179357692396e-06,
|
| 716 |
+
"loss": 0.2134,
|
| 717 |
+
"step": 1010
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 0.255,
|
| 721 |
+
"grad_norm": 1.171875,
|
| 722 |
+
"learning_rate": 4.644041074422469e-06,
|
| 723 |
+
"loss": 0.2078,
|
| 724 |
+
"step": 1020
|
| 725 |
+
},
|
| 726 |
+
{
|
| 727 |
+
"epoch": 0.2575,
|
| 728 |
+
"grad_norm": 1.484375,
|
| 729 |
+
"learning_rate": 4.632739514123733e-06,
|
| 730 |
+
"loss": 0.2229,
|
| 731 |
+
"step": 1030
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 0.26,
|
| 735 |
+
"grad_norm": 1.21875,
|
| 736 |
+
"learning_rate": 4.6212755374537596e-06,
|
| 737 |
+
"loss": 0.2166,
|
| 738 |
+
"step": 1040
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 0.2625,
|
| 742 |
+
"grad_norm": 1.3359375,
|
| 743 |
+
"learning_rate": 4.609650017438757e-06,
|
| 744 |
+
"loss": 0.2222,
|
| 745 |
+
"step": 1050
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 0.265,
|
| 749 |
+
"grad_norm": 1.640625,
|
| 750 |
+
"learning_rate": 4.5978638394070835e-06,
|
| 751 |
+
"loss": 0.2257,
|
| 752 |
+
"step": 1060
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 0.2675,
|
| 756 |
+
"grad_norm": 1.3125,
|
| 757 |
+
"learning_rate": 4.58591790092183e-06,
|
| 758 |
+
"loss": 0.2269,
|
| 759 |
+
"step": 1070
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 0.27,
|
| 763 |
+
"grad_norm": 1.140625,
|
| 764 |
+
"learning_rate": 4.5738131117124605e-06,
|
| 765 |
+
"loss": 0.2107,
|
| 766 |
+
"step": 1080
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 0.2725,
|
| 770 |
+
"grad_norm": 1.296875,
|
| 771 |
+
"learning_rate": 4.561550393605541e-06,
|
| 772 |
+
"loss": 0.2285,
|
| 773 |
+
"step": 1090
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 0.275,
|
| 777 |
+
"grad_norm": 1.2265625,
|
| 778 |
+
"learning_rate": 4.549130680454532e-06,
|
| 779 |
+
"loss": 0.219,
|
| 780 |
+
"step": 1100
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 0.2775,
|
| 784 |
+
"grad_norm": 1.328125,
|
| 785 |
+
"learning_rate": 4.536554918068673e-06,
|
| 786 |
+
"loss": 0.2203,
|
| 787 |
+
"step": 1110
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 0.28,
|
| 791 |
+
"grad_norm": 1.359375,
|
| 792 |
+
"learning_rate": 4.523824064140961e-06,
|
| 793 |
+
"loss": 0.2313,
|
| 794 |
+
"step": 1120
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 0.2825,
|
| 798 |
+
"grad_norm": 1.515625,
|
| 799 |
+
"learning_rate": 4.510939088175211e-06,
|
| 800 |
+
"loss": 0.2209,
|
| 801 |
+
"step": 1130
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 0.285,
|
| 805 |
+
"grad_norm": 1.46875,
|
| 806 |
+
"learning_rate": 4.49790097141223e-06,
|
| 807 |
+
"loss": 0.2191,
|
| 808 |
+
"step": 1140
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 0.2875,
|
| 812 |
+
"grad_norm": 1.53125,
|
| 813 |
+
"learning_rate": 4.484710706755087e-06,
|
| 814 |
+
"loss": 0.2131,
|
| 815 |
+
"step": 1150
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 0.29,
|
| 819 |
+
"grad_norm": 1.5859375,
|
| 820 |
+
"learning_rate": 4.471369298693505e-06,
|
| 821 |
+
"loss": 0.2121,
|
| 822 |
+
"step": 1160
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 0.2925,
|
| 826 |
+
"grad_norm": 1.21875,
|
| 827 |
+
"learning_rate": 4.457877763227361e-06,
|
| 828 |
+
"loss": 0.2399,
|
| 829 |
+
"step": 1170
|
| 830 |
+
},
|
| 831 |
+
{
|
| 832 |
+
"epoch": 0.295,
|
| 833 |
+
"grad_norm": 1.328125,
|
| 834 |
+
"learning_rate": 4.444237127789315e-06,
|
| 835 |
+
"loss": 0.2145,
|
| 836 |
+
"step": 1180
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 0.2975,
|
| 840 |
+
"grad_norm": 1.171875,
|
| 841 |
+
"learning_rate": 4.430448431166567e-06,
|
| 842 |
+
"loss": 0.2179,
|
| 843 |
+
"step": 1190
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 0.3,
|
| 847 |
+
"grad_norm": 1.3515625,
|
| 848 |
+
"learning_rate": 4.416512723421752e-06,
|
| 849 |
+
"loss": 0.2104,
|
| 850 |
+
"step": 1200
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"epoch": 0.3025,
|
| 854 |
+
"grad_norm": 1.234375,
|
| 855 |
+
"learning_rate": 4.402431065812968e-06,
|
| 856 |
+
"loss": 0.2194,
|
| 857 |
+
"step": 1210
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 0.305,
|
| 861 |
+
"grad_norm": 1.0546875,
|
| 862 |
+
"learning_rate": 4.388204530712959e-06,
|
| 863 |
+
"loss": 0.2179,
|
| 864 |
+
"step": 1220
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 0.3075,
|
| 868 |
+
"grad_norm": 1.1796875,
|
| 869 |
+
"learning_rate": 4.373834201527457e-06,
|
| 870 |
+
"loss": 0.2142,
|
| 871 |
+
"step": 1230
|
| 872 |
+
},
|
| 873 |
+
{
|
| 874 |
+
"epoch": 0.31,
|
| 875 |
+
"grad_norm": 1.3203125,
|
| 876 |
+
"learning_rate": 4.359321172612664e-06,
|
| 877 |
+
"loss": 0.2054,
|
| 878 |
+
"step": 1240
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 0.3125,
|
| 882 |
+
"grad_norm": 1.140625,
|
| 883 |
+
"learning_rate": 4.344666549191921e-06,
|
| 884 |
+
"loss": 0.2196,
|
| 885 |
+
"step": 1250
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 0.315,
|
| 889 |
+
"grad_norm": 1.2265625,
|
| 890 |
+
"learning_rate": 4.329871447271541e-06,
|
| 891 |
+
"loss": 0.2165,
|
| 892 |
+
"step": 1260
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"epoch": 0.3175,
|
| 896 |
+
"grad_norm": 1.3515625,
|
| 897 |
+
"learning_rate": 4.314936993555816e-06,
|
| 898 |
+
"loss": 0.2021,
|
| 899 |
+
"step": 1270
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 0.32,
|
| 903 |
+
"grad_norm": 1.453125,
|
| 904 |
+
"learning_rate": 4.299864325361217e-06,
|
| 905 |
+
"loss": 0.2113,
|
| 906 |
+
"step": 1280
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 0.3225,
|
| 910 |
+
"grad_norm": 1.375,
|
| 911 |
+
"learning_rate": 4.284654590529784e-06,
|
| 912 |
+
"loss": 0.1996,
|
| 913 |
+
"step": 1290
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 0.325,
|
| 917 |
+
"grad_norm": 1.4140625,
|
| 918 |
+
"learning_rate": 4.269308947341711e-06,
|
| 919 |
+
"loss": 0.2101,
|
| 920 |
+
"step": 1300
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 0.3275,
|
| 924 |
+
"grad_norm": 1.3515625,
|
| 925 |
+
"learning_rate": 4.25382856442714e-06,
|
| 926 |
+
"loss": 0.2086,
|
| 927 |
+
"step": 1310
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 0.33,
|
| 931 |
+
"grad_norm": 1.453125,
|
| 932 |
+
"learning_rate": 4.238214620677164e-06,
|
| 933 |
+
"loss": 0.2125,
|
| 934 |
+
"step": 1320
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 0.3325,
|
| 938 |
+
"grad_norm": 1.1796875,
|
| 939 |
+
"learning_rate": 4.222468305154052e-06,
|
| 940 |
+
"loss": 0.2162,
|
| 941 |
+
"step": 1330
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 0.335,
|
| 945 |
+
"grad_norm": 1.25,
|
| 946 |
+
"learning_rate": 4.206590817000695e-06,
|
| 947 |
+
"loss": 0.2127,
|
| 948 |
+
"step": 1340
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 0.3375,
|
| 952 |
+
"grad_norm": 1.1796875,
|
| 953 |
+
"learning_rate": 4.190583365349289e-06,
|
| 954 |
+
"loss": 0.2095,
|
| 955 |
+
"step": 1350
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 0.34,
|
| 959 |
+
"grad_norm": 1.2578125,
|
| 960 |
+
"learning_rate": 4.174447169229252e-06,
|
| 961 |
+
"loss": 0.217,
|
| 962 |
+
"step": 1360
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 0.3425,
|
| 966 |
+
"grad_norm": 1.4140625,
|
| 967 |
+
"learning_rate": 4.158183457474392e-06,
|
| 968 |
+
"loss": 0.212,
|
| 969 |
+
"step": 1370
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 0.345,
|
| 973 |
+
"grad_norm": 1.4453125,
|
| 974 |
+
"learning_rate": 4.141793468629327e-06,
|
| 975 |
+
"loss": 0.2221,
|
| 976 |
+
"step": 1380
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"epoch": 0.3475,
|
| 980 |
+
"grad_norm": 1.125,
|
| 981 |
+
"learning_rate": 4.125278450855165e-06,
|
| 982 |
+
"loss": 0.2079,
|
| 983 |
+
"step": 1390
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 0.35,
|
| 987 |
+
"grad_norm": 1.28125,
|
| 988 |
+
"learning_rate": 4.1086396618344474e-06,
|
| 989 |
+
"loss": 0.2085,
|
| 990 |
+
"step": 1400
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 0.3525,
|
| 994 |
+
"grad_norm": 1.3828125,
|
| 995 |
+
"learning_rate": 4.09187836867538e-06,
|
| 996 |
+
"loss": 0.2123,
|
| 997 |
+
"step": 1410
|
| 998 |
+
},
|
| 999 |
+
{
|
| 1000 |
+
"epoch": 0.355,
|
| 1001 |
+
"grad_norm": 1.21875,
|
| 1002 |
+
"learning_rate": 4.074995847815331e-06,
|
| 1003 |
+
"loss": 0.2116,
|
| 1004 |
+
"step": 1420
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 0.3575,
|
| 1008 |
+
"grad_norm": 1.3671875,
|
| 1009 |
+
"learning_rate": 4.057993384923626e-06,
|
| 1010 |
+
"loss": 0.2288,
|
| 1011 |
+
"step": 1430
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 0.36,
|
| 1015 |
+
"grad_norm": 1.1484375,
|
| 1016 |
+
"learning_rate": 4.0408722748036426e-06,
|
| 1017 |
+
"loss": 0.2038,
|
| 1018 |
+
"step": 1440
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 0.3625,
|
| 1022 |
+
"grad_norm": 1.28125,
|
| 1023 |
+
"learning_rate": 4.023633821294203e-06,
|
| 1024 |
+
"loss": 0.2123,
|
| 1025 |
+
"step": 1450
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 0.365,
|
| 1029 |
+
"grad_norm": 1.03125,
|
| 1030 |
+
"learning_rate": 4.006279337170283e-06,
|
| 1031 |
+
"loss": 0.1979,
|
| 1032 |
+
"step": 1460
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 0.3675,
|
| 1036 |
+
"grad_norm": 1.2109375,
|
| 1037 |
+
"learning_rate": 3.988810144043041e-06,
|
| 1038 |
+
"loss": 0.2113,
|
| 1039 |
+
"step": 1470
|
| 1040 |
+
},
|
| 1041 |
+
{
|
| 1042 |
+
"epoch": 0.37,
|
| 1043 |
+
"grad_norm": 1.4453125,
|
| 1044 |
+
"learning_rate": 3.971227572259167e-06,
|
| 1045 |
+
"loss": 0.2077,
|
| 1046 |
+
"step": 1480
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 0.3725,
|
| 1050 |
+
"grad_norm": 1.1328125,
|
| 1051 |
+
"learning_rate": 3.953532960799577e-06,
|
| 1052 |
+
"loss": 0.1986,
|
| 1053 |
+
"step": 1490
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 0.375,
|
| 1057 |
+
"grad_norm": 1.1484375,
|
| 1058 |
+
"learning_rate": 3.935727657177439e-06,
|
| 1059 |
+
"loss": 0.1972,
|
| 1060 |
+
"step": 1500
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"epoch": 0.3775,
|
| 1064 |
+
"grad_norm": 1.40625,
|
| 1065 |
+
"learning_rate": 3.917813017335562e-06,
|
| 1066 |
+
"loss": 0.2082,
|
| 1067 |
+
"step": 1510
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 0.38,
|
| 1071 |
+
"grad_norm": 1.375,
|
| 1072 |
+
"learning_rate": 3.899790405543129e-06,
|
| 1073 |
+
"loss": 0.2086,
|
| 1074 |
+
"step": 1520
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 0.3825,
|
| 1078 |
+
"grad_norm": 1.2734375,
|
| 1079 |
+
"learning_rate": 3.881661194291805e-06,
|
| 1080 |
+
"loss": 0.1968,
|
| 1081 |
+
"step": 1530
|
| 1082 |
+
},
|
| 1083 |
+
{
|
| 1084 |
+
"epoch": 0.385,
|
| 1085 |
+
"grad_norm": 1.1640625,
|
| 1086 |
+
"learning_rate": 3.863426764191216e-06,
|
| 1087 |
+
"loss": 0.2064,
|
| 1088 |
+
"step": 1540
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 0.3875,
|
| 1092 |
+
"grad_norm": 1.1015625,
|
| 1093 |
+
"learning_rate": 3.845088503863813e-06,
|
| 1094 |
+
"loss": 0.1855,
|
| 1095 |
+
"step": 1550
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 0.39,
|
| 1099 |
+
"grad_norm": 1.2109375,
|
| 1100 |
+
"learning_rate": 3.826647809839119e-06,
|
| 1101 |
+
"loss": 0.2039,
|
| 1102 |
+
"step": 1560
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"epoch": 0.3925,
|
| 1106 |
+
"grad_norm": 1.125,
|
| 1107 |
+
"learning_rate": 3.8081060864473794e-06,
|
| 1108 |
+
"loss": 0.2079,
|
| 1109 |
+
"step": 1570
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 0.395,
|
| 1113 |
+
"grad_norm": 1.09375,
|
| 1114 |
+
"learning_rate": 3.7894647457126188e-06,
|
| 1115 |
+
"loss": 0.2126,
|
| 1116 |
+
"step": 1580
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 0.3975,
|
| 1120 |
+
"grad_norm": 1.2734375,
|
| 1121 |
+
"learning_rate": 3.770725207245106e-06,
|
| 1122 |
+
"loss": 0.2156,
|
| 1123 |
+
"step": 1590
|
| 1124 |
+
},
|
| 1125 |
+
{
|
| 1126 |
+
"epoch": 0.4,
|
| 1127 |
+
"grad_norm": 1.1015625,
|
| 1128 |
+
"learning_rate": 3.751888898133249e-06,
|
| 1129 |
+
"loss": 0.2122,
|
| 1130 |
+
"step": 1600
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 0.4025,
|
| 1134 |
+
"grad_norm": 1.3046875,
|
| 1135 |
+
"learning_rate": 3.7329572528349145e-06,
|
| 1136 |
+
"loss": 0.1996,
|
| 1137 |
+
"step": 1610
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 0.405,
|
| 1141 |
+
"grad_norm": 1.2734375,
|
| 1142 |
+
"learning_rate": 3.7139317130681886e-06,
|
| 1143 |
+
"loss": 0.2136,
|
| 1144 |
+
"step": 1620
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"epoch": 0.4075,
|
| 1148 |
+
"grad_norm": 0.98046875,
|
| 1149 |
+
"learning_rate": 3.694813727701584e-06,
|
| 1150 |
+
"loss": 0.2055,
|
| 1151 |
+
"step": 1630
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 0.41,
|
| 1155 |
+
"grad_norm": 1.421875,
|
| 1156 |
+
"learning_rate": 3.675604752643706e-06,
|
| 1157 |
+
"loss": 0.2184,
|
| 1158 |
+
"step": 1640
|
| 1159 |
+
},
|
| 1160 |
+
{
|
| 1161 |
+
"epoch": 0.4125,
|
| 1162 |
+
"grad_norm": 1.2265625,
|
| 1163 |
+
"learning_rate": 3.6563062507323752e-06,
|
| 1164 |
+
"loss": 0.2068,
|
| 1165 |
+
"step": 1650
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 0.415,
|
| 1169 |
+
"grad_norm": 1.09375,
|
| 1170 |
+
"learning_rate": 3.6369196916232297e-06,
|
| 1171 |
+
"loss": 0.2034,
|
| 1172 |
+
"step": 1660
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 0.4175,
|
| 1176 |
+
"grad_norm": 1.1640625,
|
| 1177 |
+
"learning_rate": 3.6174465516778032e-06,
|
| 1178 |
+
"loss": 0.1992,
|
| 1179 |
+
"step": 1670
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 0.42,
|
| 1183 |
+
"grad_norm": 1.1171875,
|
| 1184 |
+
"learning_rate": 3.5978883138510963e-06,
|
| 1185 |
+
"loss": 0.2037,
|
| 1186 |
+
"step": 1680
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 0.4225,
|
| 1190 |
+
"grad_norm": 1.15625,
|
| 1191 |
+
"learning_rate": 3.578246467578642e-06,
|
| 1192 |
+
"loss": 0.2185,
|
| 1193 |
+
"step": 1690
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 0.425,
|
| 1197 |
+
"grad_norm": 1.1875,
|
| 1198 |
+
"learning_rate": 3.558522508663081e-06,
|
| 1199 |
+
"loss": 0.2038,
|
| 1200 |
+
"step": 1700
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 0.4275,
|
| 1204 |
+
"grad_norm": 1.4296875,
|
| 1205 |
+
"learning_rate": 3.538717939160249e-06,
|
| 1206 |
+
"loss": 0.2088,
|
| 1207 |
+
"step": 1710
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 0.43,
|
| 1211 |
+
"grad_norm": 1.453125,
|
| 1212 |
+
"learning_rate": 3.5188342672647897e-06,
|
| 1213 |
+
"loss": 0.2041,
|
| 1214 |
+
"step": 1720
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 0.4325,
|
| 1218 |
+
"grad_norm": 1.125,
|
| 1219 |
+
"learning_rate": 3.4988730071953005e-06,
|
| 1220 |
+
"loss": 0.2065,
|
| 1221 |
+
"step": 1730
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 0.435,
|
| 1225 |
+
"grad_norm": 1.171875,
|
| 1226 |
+
"learning_rate": 3.478835679079019e-06,
|
| 1227 |
+
"loss": 0.2041,
|
| 1228 |
+
"step": 1740
|
| 1229 |
+
},
|
| 1230 |
+
{
|
| 1231 |
+
"epoch": 0.4375,
|
| 1232 |
+
"grad_norm": 1.5390625,
|
| 1233 |
+
"learning_rate": 3.4587238088360605e-06,
|
| 1234 |
+
"loss": 0.1996,
|
| 1235 |
+
"step": 1750
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 0.44,
|
| 1239 |
+
"grad_norm": 1.390625,
|
| 1240 |
+
"learning_rate": 3.438538928063208e-06,
|
| 1241 |
+
"loss": 0.1931,
|
| 1242 |
+
"step": 1760
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 0.4425,
|
| 1246 |
+
"grad_norm": 1.2265625,
|
| 1247 |
+
"learning_rate": 3.4182825739172826e-06,
|
| 1248 |
+
"loss": 0.2095,
|
| 1249 |
+
"step": 1770
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 0.445,
|
| 1253 |
+
"grad_norm": 1.140625,
|
| 1254 |
+
"learning_rate": 3.3979562889980777e-06,
|
| 1255 |
+
"loss": 0.2097,
|
| 1256 |
+
"step": 1780
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 0.4475,
|
| 1260 |
+
"grad_norm": 1.34375,
|
| 1261 |
+
"learning_rate": 3.377561621230887e-06,
|
| 1262 |
+
"loss": 0.1874,
|
| 1263 |
+
"step": 1790
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 0.45,
|
| 1267 |
+
"grad_norm": 1.171875,
|
| 1268 |
+
"learning_rate": 3.357100123748621e-06,
|
| 1269 |
+
"loss": 0.2025,
|
| 1270 |
+
"step": 1800
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 0.4525,
|
| 1274 |
+
"grad_norm": 0.98828125,
|
| 1275 |
+
"learning_rate": 3.3365733547735334e-06,
|
| 1276 |
+
"loss": 0.1846,
|
| 1277 |
+
"step": 1810
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 0.455,
|
| 1281 |
+
"grad_norm": 1.28125,
|
| 1282 |
+
"learning_rate": 3.315982877498555e-06,
|
| 1283 |
+
"loss": 0.2049,
|
| 1284 |
+
"step": 1820
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 0.4575,
|
| 1288 |
+
"grad_norm": 1.1640625,
|
| 1289 |
+
"learning_rate": 3.2953302599682487e-06,
|
| 1290 |
+
"loss": 0.2021,
|
| 1291 |
+
"step": 1830
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 0.46,
|
| 1295 |
+
"grad_norm": 1.265625,
|
| 1296 |
+
"learning_rate": 3.2746170749593998e-06,
|
| 1297 |
+
"loss": 0.1979,
|
| 1298 |
+
"step": 1840
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 0.4625,
|
| 1302 |
+
"grad_norm": 1.3671875,
|
| 1303 |
+
"learning_rate": 3.2538448998612394e-06,
|
| 1304 |
+
"loss": 0.1965,
|
| 1305 |
+
"step": 1850
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 0.465,
|
| 1309 |
+
"grad_norm": 1.1328125,
|
| 1310 |
+
"learning_rate": 3.233015316555326e-06,
|
| 1311 |
+
"loss": 0.2151,
|
| 1312 |
+
"step": 1860
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 0.4675,
|
| 1316 |
+
"grad_norm": 1.078125,
|
| 1317 |
+
"learning_rate": 3.212129911295074e-06,
|
| 1318 |
+
"loss": 0.1824,
|
| 1319 |
+
"step": 1870
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 0.47,
|
| 1323 |
+
"grad_norm": 1.109375,
|
| 1324 |
+
"learning_rate": 3.1911902745849526e-06,
|
| 1325 |
+
"loss": 0.2051,
|
| 1326 |
+
"step": 1880
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 0.4725,
|
| 1330 |
+
"grad_norm": 1.09375,
|
| 1331 |
+
"learning_rate": 3.17019800105937e-06,
|
| 1332 |
+
"loss": 0.2145,
|
| 1333 |
+
"step": 1890
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 0.475,
|
| 1337 |
+
"grad_norm": 1.125,
|
| 1338 |
+
"learning_rate": 3.1491546893612296e-06,
|
| 1339 |
+
"loss": 0.2003,
|
| 1340 |
+
"step": 1900
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 0.4775,
|
| 1344 |
+
"grad_norm": 1.2109375,
|
| 1345 |
+
"learning_rate": 3.128061942020189e-06,
|
| 1346 |
+
"loss": 0.2078,
|
| 1347 |
+
"step": 1910
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 0.48,
|
| 1351 |
+
"grad_norm": 1.3671875,
|
| 1352 |
+
"learning_rate": 3.1069213653306242e-06,
|
| 1353 |
+
"loss": 0.1979,
|
| 1354 |
+
"step": 1920
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 0.4825,
|
| 1358 |
+
"grad_norm": 1.109375,
|
| 1359 |
+
"learning_rate": 3.0857345692292968e-06,
|
| 1360 |
+
"loss": 0.1925,
|
| 1361 |
+
"step": 1930
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 0.485,
|
| 1365 |
+
"grad_norm": 1.171875,
|
| 1366 |
+
"learning_rate": 3.0645031671727598e-06,
|
| 1367 |
+
"loss": 0.2042,
|
| 1368 |
+
"step": 1940
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 0.4875,
|
| 1372 |
+
"grad_norm": 1.1875,
|
| 1373 |
+
"learning_rate": 3.0432287760144797e-06,
|
| 1374 |
+
"loss": 0.2064,
|
| 1375 |
+
"step": 1950
|
| 1376 |
+
},
|
| 1377 |
+
{
|
| 1378 |
+
"epoch": 0.49,
|
| 1379 |
+
"grad_norm": 1.4921875,
|
| 1380 |
+
"learning_rate": 3.0219130158817093e-06,
|
| 1381 |
+
"loss": 0.2035,
|
| 1382 |
+
"step": 1960
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 0.4925,
|
| 1386 |
+
"grad_norm": 1.125,
|
| 1387 |
+
"learning_rate": 3.0005575100521115e-06,
|
| 1388 |
+
"loss": 0.195,
|
| 1389 |
+
"step": 1970
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 0.495,
|
| 1393 |
+
"grad_norm": 1.3515625,
|
| 1394 |
+
"learning_rate": 2.979163884830137e-06,
|
| 1395 |
+
"loss": 0.2044,
|
| 1396 |
+
"step": 1980
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 0.4975,
|
| 1400 |
+
"grad_norm": 1.765625,
|
| 1401 |
+
"learning_rate": 2.957733769423174e-06,
|
| 1402 |
+
"loss": 0.212,
|
| 1403 |
+
"step": 1990
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 0.5,
|
| 1407 |
+
"grad_norm": 1.28125,
|
| 1408 |
+
"learning_rate": 2.93626879581748e-06,
|
| 1409 |
+
"loss": 0.1985,
|
| 1410 |
+
"step": 2000
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 0.5025,
|
| 1414 |
+
"grad_norm": 1.359375,
|
| 1415 |
+
"learning_rate": 2.914770598653902e-06,
|
| 1416 |
+
"loss": 0.1872,
|
| 1417 |
+
"step": 2010
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 0.505,
|
| 1421 |
+
"grad_norm": 1.1328125,
|
| 1422 |
+
"learning_rate": 2.8932408151033868e-06,
|
| 1423 |
+
"loss": 0.1884,
|
| 1424 |
+
"step": 2020
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 0.5075,
|
| 1428 |
+
"grad_norm": 1.203125,
|
| 1429 |
+
"learning_rate": 2.8716810847423083e-06,
|
| 1430 |
+
"loss": 0.2003,
|
| 1431 |
+
"step": 2030
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 0.51,
|
| 1435 |
+
"grad_norm": 1.2265625,
|
| 1436 |
+
"learning_rate": 2.8500930494276035e-06,
|
| 1437 |
+
"loss": 0.2057,
|
| 1438 |
+
"step": 2040
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"epoch": 0.5125,
|
| 1442 |
+
"grad_norm": 1.2890625,
|
| 1443 |
+
"learning_rate": 2.828478353171745e-06,
|
| 1444 |
+
"loss": 0.197,
|
| 1445 |
+
"step": 2050
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 0.515,
|
| 1449 |
+
"grad_norm": 1.2890625,
|
| 1450 |
+
"learning_rate": 2.8068386420175376e-06,
|
| 1451 |
+
"loss": 0.1982,
|
| 1452 |
+
"step": 2060
|
| 1453 |
+
},
|
| 1454 |
+
{
|
| 1455 |
+
"epoch": 0.5175,
|
| 1456 |
+
"grad_norm": 1.0390625,
|
| 1457 |
+
"learning_rate": 2.785175563912766e-06,
|
| 1458 |
+
"loss": 0.1865,
|
| 1459 |
+
"step": 2070
|
| 1460 |
+
},
|
| 1461 |
+
{
|
| 1462 |
+
"epoch": 0.52,
|
| 1463 |
+
"grad_norm": 1.1171875,
|
| 1464 |
+
"learning_rate": 2.7634907685846995e-06,
|
| 1465 |
+
"loss": 0.1963,
|
| 1466 |
+
"step": 2080
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 0.5225,
|
| 1470 |
+
"grad_norm": 1.3515625,
|
| 1471 |
+
"learning_rate": 2.7417859074144604e-06,
|
| 1472 |
+
"loss": 0.1976,
|
| 1473 |
+
"step": 2090
|
| 1474 |
+
},
|
| 1475 |
+
{
|
| 1476 |
+
"epoch": 0.525,
|
| 1477 |
+
"grad_norm": 1.234375,
|
| 1478 |
+
"learning_rate": 2.7200626333112595e-06,
|
| 1479 |
+
"loss": 0.1997,
|
| 1480 |
+
"step": 2100
|
| 1481 |
+
},
|
| 1482 |
+
{
|
| 1483 |
+
"epoch": 0.5275,
|
| 1484 |
+
"grad_norm": 1.0703125,
|
| 1485 |
+
"learning_rate": 2.6983226005865236e-06,
|
| 1486 |
+
"loss": 0.1997,
|
| 1487 |
+
"step": 2110
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 0.53,
|
| 1491 |
+
"grad_norm": 1.046875,
|
| 1492 |
+
"learning_rate": 2.676567464827917e-06,
|
| 1493 |
+
"loss": 0.2114,
|
| 1494 |
+
"step": 2120
|
| 1495 |
+
},
|
| 1496 |
+
{
|
| 1497 |
+
"epoch": 0.5325,
|
| 1498 |
+
"grad_norm": 1.1953125,
|
| 1499 |
+
"learning_rate": 2.6547988827732546e-06,
|
| 1500 |
+
"loss": 0.194,
|
| 1501 |
+
"step": 2130
|
| 1502 |
+
},
|
| 1503 |
+
{
|
| 1504 |
+
"epoch": 0.535,
|
| 1505 |
+
"grad_norm": 1.1015625,
|
| 1506 |
+
"learning_rate": 2.633018512184341e-06,
|
| 1507 |
+
"loss": 0.198,
|
| 1508 |
+
"step": 2140
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 0.5375,
|
| 1512 |
+
"grad_norm": 1.2265625,
|
| 1513 |
+
"learning_rate": 2.6112280117207223e-06,
|
| 1514 |
+
"loss": 0.2126,
|
| 1515 |
+
"step": 2150
|
| 1516 |
+
},
|
| 1517 |
+
{
|
| 1518 |
+
"epoch": 0.54,
|
| 1519 |
+
"grad_norm": 1.40625,
|
| 1520 |
+
"learning_rate": 2.5894290408133744e-06,
|
| 1521 |
+
"loss": 0.2021,
|
| 1522 |
+
"step": 2160
|
| 1523 |
+
},
|
| 1524 |
+
{
|
| 1525 |
+
"epoch": 0.5425,
|
| 1526 |
+
"grad_norm": 1.2890625,
|
| 1527 |
+
"learning_rate": 2.56762325953833e-06,
|
| 1528 |
+
"loss": 0.1997,
|
| 1529 |
+
"step": 2170
|
| 1530 |
+
},
|
| 1531 |
+
{
|
| 1532 |
+
"epoch": 0.545,
|
| 1533 |
+
"grad_norm": 1.140625,
|
| 1534 |
+
"learning_rate": 2.5458123284902577e-06,
|
| 1535 |
+
"loss": 0.1984,
|
| 1536 |
+
"step": 2180
|
| 1537 |
+
},
|
| 1538 |
+
{
|
| 1539 |
+
"epoch": 0.5475,
|
| 1540 |
+
"grad_norm": 1.0078125,
|
| 1541 |
+
"learning_rate": 2.5239979086560003e-06,
|
| 1542 |
+
"loss": 0.1981,
|
| 1543 |
+
"step": 2190
|
| 1544 |
+
},
|
| 1545 |
+
{
|
| 1546 |
+
"epoch": 0.55,
|
| 1547 |
+
"grad_norm": 1.2890625,
|
| 1548 |
+
"learning_rate": 2.5021816612880884e-06,
|
| 1549 |
+
"loss": 0.2044,
|
| 1550 |
+
"step": 2200
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 0.5525,
|
| 1554 |
+
"grad_norm": 1.1953125,
|
| 1555 |
+
"learning_rate": 2.4803652477782228e-06,
|
| 1556 |
+
"loss": 0.2039,
|
| 1557 |
+
"step": 2210
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 0.555,
|
| 1561 |
+
"grad_norm": 1.078125,
|
| 1562 |
+
"learning_rate": 2.4585503295307565e-06,
|
| 1563 |
+
"loss": 0.203,
|
| 1564 |
+
"step": 2220
|
| 1565 |
+
},
|
| 1566 |
+
{
|
| 1567 |
+
"epoch": 0.5575,
|
| 1568 |
+
"grad_norm": 1.109375,
|
| 1569 |
+
"learning_rate": 2.436738567836176e-06,
|
| 1570 |
+
"loss": 0.1996,
|
| 1571 |
+
"step": 2230
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"epoch": 0.56,
|
| 1575 |
+
"grad_norm": 1.09375,
|
| 1576 |
+
"learning_rate": 2.4149316237445813e-06,
|
| 1577 |
+
"loss": 0.1979,
|
| 1578 |
+
"step": 2240
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 0.5625,
|
| 1582 |
+
"grad_norm": 1.078125,
|
| 1583 |
+
"learning_rate": 2.3931311579391946e-06,
|
| 1584 |
+
"loss": 0.1939,
|
| 1585 |
+
"step": 2250
|
| 1586 |
+
},
|
| 1587 |
+
{
|
| 1588 |
+
"epoch": 0.565,
|
| 1589 |
+
"grad_norm": 1.3359375,
|
| 1590 |
+
"learning_rate": 2.37133883060989e-06,
|
| 1591 |
+
"loss": 0.1991,
|
| 1592 |
+
"step": 2260
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 0.5675,
|
| 1596 |
+
"grad_norm": 1.140625,
|
| 1597 |
+
"learning_rate": 2.3495563013267668e-06,
|
| 1598 |
+
"loss": 0.19,
|
| 1599 |
+
"step": 2270
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 0.57,
|
| 1603 |
+
"grad_norm": 1.125,
|
| 1604 |
+
"learning_rate": 2.3277852289137636e-06,
|
| 1605 |
+
"loss": 0.1957,
|
| 1606 |
+
"step": 2280
|
| 1607 |
+
},
|
| 1608 |
+
{
|
| 1609 |
+
"epoch": 0.5725,
|
| 1610 |
+
"grad_norm": 1.2578125,
|
| 1611 |
+
"learning_rate": 2.306027271322336e-06,
|
| 1612 |
+
"loss": 0.1966,
|
| 1613 |
+
"step": 2290
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"epoch": 0.575,
|
| 1617 |
+
"grad_norm": 1.203125,
|
| 1618 |
+
"learning_rate": 2.284284085505192e-06,
|
| 1619 |
+
"loss": 0.2099,
|
| 1620 |
+
"step": 2300
|
| 1621 |
+
},
|
| 1622 |
+
{
|
| 1623 |
+
"epoch": 0.5775,
|
| 1624 |
+
"grad_norm": 1.2265625,
|
| 1625 |
+
"learning_rate": 2.2625573272901156e-06,
|
| 1626 |
+
"loss": 0.1997,
|
| 1627 |
+
"step": 2310
|
| 1628 |
+
},
|
| 1629 |
+
{
|
| 1630 |
+
"epoch": 0.58,
|
| 1631 |
+
"grad_norm": 1.1796875,
|
| 1632 |
+
"learning_rate": 2.240848651253863e-06,
|
| 1633 |
+
"loss": 0.1897,
|
| 1634 |
+
"step": 2320
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 0.5825,
|
| 1638 |
+
"grad_norm": 0.9921875,
|
| 1639 |
+
"learning_rate": 2.2191597105961613e-06,
|
| 1640 |
+
"loss": 0.1826,
|
| 1641 |
+
"step": 2330
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 0.585,
|
| 1645 |
+
"grad_norm": 1.109375,
|
| 1646 |
+
"learning_rate": 2.1974921570138155e-06,
|
| 1647 |
+
"loss": 0.1917,
|
| 1648 |
+
"step": 2340
|
| 1649 |
+
},
|
| 1650 |
+
{
|
| 1651 |
+
"epoch": 0.5875,
|
| 1652 |
+
"grad_norm": 1.265625,
|
| 1653 |
+
"learning_rate": 2.1758476405749207e-06,
|
| 1654 |
+
"loss": 0.1963,
|
| 1655 |
+
"step": 2350
|
| 1656 |
+
},
|
| 1657 |
+
{
|
| 1658 |
+
"epoch": 0.59,
|
| 1659 |
+
"grad_norm": 1.140625,
|
| 1660 |
+
"learning_rate": 2.154227809593203e-06,
|
| 1661 |
+
"loss": 0.1931,
|
| 1662 |
+
"step": 2360
|
| 1663 |
+
},
|
| 1664 |
+
{
|
| 1665 |
+
"epoch": 0.5925,
|
| 1666 |
+
"grad_norm": 1.1953125,
|
| 1667 |
+
"learning_rate": 2.1326343105024962e-06,
|
| 1668 |
+
"loss": 0.202,
|
| 1669 |
+
"step": 2370
|
| 1670 |
+
},
|
| 1671 |
+
{
|
| 1672 |
+
"epoch": 0.595,
|
| 1673 |
+
"grad_norm": 1.2265625,
|
| 1674 |
+
"learning_rate": 2.111068787731358e-06,
|
| 1675 |
+
"loss": 0.2061,
|
| 1676 |
+
"step": 2380
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 0.5975,
|
| 1680 |
+
"grad_norm": 1.578125,
|
| 1681 |
+
"learning_rate": 2.089532883577843e-06,
|
| 1682 |
+
"loss": 0.1867,
|
| 1683 |
+
"step": 2390
|
| 1684 |
+
},
|
| 1685 |
+
{
|
| 1686 |
+
"epoch": 0.6,
|
| 1687 |
+
"grad_norm": 1.234375,
|
| 1688 |
+
"learning_rate": 2.068028238084432e-06,
|
| 1689 |
+
"loss": 0.2039,
|
| 1690 |
+
"step": 2400
|
| 1691 |
+
},
|
| 1692 |
+
{
|
| 1693 |
+
"epoch": 0.6025,
|
| 1694 |
+
"grad_norm": 1.171875,
|
| 1695 |
+
"learning_rate": 2.046556488913137e-06,
|
| 1696 |
+
"loss": 0.209,
|
| 1697 |
+
"step": 2410
|
| 1698 |
+
},
|
| 1699 |
+
{
|
| 1700 |
+
"epoch": 0.605,
|
| 1701 |
+
"grad_norm": 1.234375,
|
| 1702 |
+
"learning_rate": 2.025119271220789e-06,
|
| 1703 |
+
"loss": 0.1894,
|
| 1704 |
+
"step": 2420
|
| 1705 |
+
},
|
| 1706 |
+
{
|
| 1707 |
+
"epoch": 0.6075,
|
| 1708 |
+
"grad_norm": 1.125,
|
| 1709 |
+
"learning_rate": 2.0037182175345137e-06,
|
| 1710 |
+
"loss": 0.2098,
|
| 1711 |
+
"step": 2430
|
| 1712 |
+
},
|
| 1713 |
+
{
|
| 1714 |
+
"epoch": 0.61,
|
| 1715 |
+
"grad_norm": 1.1171875,
|
| 1716 |
+
"learning_rate": 1.9823549576274048e-06,
|
| 1717 |
+
"loss": 0.1955,
|
| 1718 |
+
"step": 2440
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 0.6125,
|
| 1722 |
+
"grad_norm": 1.1640625,
|
| 1723 |
+
"learning_rate": 1.961031118394418e-06,
|
| 1724 |
+
"loss": 0.1881,
|
| 1725 |
+
"step": 2450
|
| 1726 |
+
},
|
| 1727 |
+
{
|
| 1728 |
+
"epoch": 0.615,
|
| 1729 |
+
"grad_norm": 1.453125,
|
| 1730 |
+
"learning_rate": 1.939748323728468e-06,
|
| 1731 |
+
"loss": 0.1976,
|
| 1732 |
+
"step": 2460
|
| 1733 |
+
},
|
| 1734 |
+
{
|
| 1735 |
+
"epoch": 0.6175,
|
| 1736 |
+
"grad_norm": 1.4296875,
|
| 1737 |
+
"learning_rate": 1.918508194396769e-06,
|
| 1738 |
+
"loss": 0.2028,
|
| 1739 |
+
"step": 2470
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 0.62,
|
| 1743 |
+
"grad_norm": 1.828125,
|
| 1744 |
+
"learning_rate": 1.8973123479174038e-06,
|
| 1745 |
+
"loss": 0.2065,
|
| 1746 |
+
"step": 2480
|
| 1747 |
+
},
|
| 1748 |
+
{
|
| 1749 |
+
"epoch": 0.6225,
|
| 1750 |
+
"grad_norm": 1.3359375,
|
| 1751 |
+
"learning_rate": 1.8761623984361444e-06,
|
| 1752 |
+
"loss": 0.198,
|
| 1753 |
+
"step": 2490
|
| 1754 |
+
},
|
| 1755 |
+
{
|
| 1756 |
+
"epoch": 0.625,
|
| 1757 |
+
"grad_norm": 1.1171875,
|
| 1758 |
+
"learning_rate": 1.8550599566035299e-06,
|
| 1759 |
+
"loss": 0.192,
|
| 1760 |
+
"step": 2500
|
| 1761 |
+
},
|
| 1762 |
+
{
|
| 1763 |
+
"epoch": 0.6275,
|
| 1764 |
+
"grad_norm": 1.2265625,
|
| 1765 |
+
"learning_rate": 1.834006629452207e-06,
|
| 1766 |
+
"loss": 0.1894,
|
| 1767 |
+
"step": 2510
|
| 1768 |
+
},
|
| 1769 |
+
{
|
| 1770 |
+
"epoch": 0.63,
|
| 1771 |
+
"grad_norm": 1.15625,
|
| 1772 |
+
"learning_rate": 1.8130040202745488e-06,
|
| 1773 |
+
"loss": 0.203,
|
| 1774 |
+
"step": 2520
|
| 1775 |
+
},
|
| 1776 |
+
{
|
| 1777 |
+
"epoch": 0.6325,
|
| 1778 |
+
"grad_norm": 1.171875,
|
| 1779 |
+
"learning_rate": 1.7920537285005607e-06,
|
| 1780 |
+
"loss": 0.1879,
|
| 1781 |
+
"step": 2530
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 0.635,
|
| 1785 |
+
"grad_norm": 1.1953125,
|
| 1786 |
+
"learning_rate": 1.7711573495760725e-06,
|
| 1787 |
+
"loss": 0.1887,
|
| 1788 |
+
"step": 2540
|
| 1789 |
+
},
|
| 1790 |
+
{
|
| 1791 |
+
"epoch": 0.6375,
|
| 1792 |
+
"grad_norm": 1.2578125,
|
| 1793 |
+
"learning_rate": 1.750316474841242e-06,
|
| 1794 |
+
"loss": 0.1894,
|
| 1795 |
+
"step": 2550
|
| 1796 |
+
},
|
| 1797 |
+
{
|
| 1798 |
+
"epoch": 0.64,
|
| 1799 |
+
"grad_norm": 1.1015625,
|
| 1800 |
+
"learning_rate": 1.7295326914093713e-06,
|
| 1801 |
+
"loss": 0.1991,
|
| 1802 |
+
"step": 2560
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 0.6425,
|
| 1806 |
+
"grad_norm": 1.3984375,
|
| 1807 |
+
"learning_rate": 1.7088075820460348e-06,
|
| 1808 |
+
"loss": 0.1982,
|
| 1809 |
+
"step": 2570
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 0.645,
|
| 1813 |
+
"grad_norm": 1.1171875,
|
| 1814 |
+
"learning_rate": 1.6881427250485516e-06,
|
| 1815 |
+
"loss": 0.1876,
|
| 1816 |
+
"step": 2580
|
| 1817 |
+
},
|
| 1818 |
+
{
|
| 1819 |
+
"epoch": 0.6475,
|
| 1820 |
+
"grad_norm": 1.1015625,
|
| 1821 |
+
"learning_rate": 1.6675396941257896e-06,
|
| 1822 |
+
"loss": 0.1853,
|
| 1823 |
+
"step": 2590
|
| 1824 |
+
},
|
| 1825 |
+
{
|
| 1826 |
+
"epoch": 0.65,
|
| 1827 |
+
"grad_norm": 1.1171875,
|
| 1828 |
+
"learning_rate": 1.6470000582783205e-06,
|
| 1829 |
+
"loss": 0.1917,
|
| 1830 |
+
"step": 2600
|
| 1831 |
+
},
|
| 1832 |
+
{
|
| 1833 |
+
"epoch": 0.6525,
|
| 1834 |
+
"grad_norm": 1.296875,
|
| 1835 |
+
"learning_rate": 1.6265253816789372e-06,
|
| 1836 |
+
"loss": 0.2017,
|
| 1837 |
+
"step": 2610
|
| 1838 |
+
},
|
| 1839 |
+
{
|
| 1840 |
+
"epoch": 0.655,
|
| 1841 |
+
"grad_norm": 1.203125,
|
| 1842 |
+
"learning_rate": 1.6061172235535342e-06,
|
| 1843 |
+
"loss": 0.193,
|
| 1844 |
+
"step": 2620
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 0.6575,
|
| 1848 |
+
"grad_norm": 1.25,
|
| 1849 |
+
"learning_rate": 1.5857771380623643e-06,
|
| 1850 |
+
"loss": 0.2018,
|
| 1851 |
+
"step": 2630
|
| 1852 |
+
},
|
| 1853 |
+
{
|
| 1854 |
+
"epoch": 0.66,
|
| 1855 |
+
"grad_norm": 1.4765625,
|
| 1856 |
+
"learning_rate": 1.5655066741816898e-06,
|
| 1857 |
+
"loss": 0.2062,
|
| 1858 |
+
"step": 2640
|
| 1859 |
+
},
|
| 1860 |
+
{
|
| 1861 |
+
"epoch": 0.6625,
|
| 1862 |
+
"grad_norm": 1.1484375,
|
| 1863 |
+
"learning_rate": 1.545307375585814e-06,
|
| 1864 |
+
"loss": 0.1961,
|
| 1865 |
+
"step": 2650
|
| 1866 |
+
},
|
| 1867 |
+
{
|
| 1868 |
+
"epoch": 0.665,
|
| 1869 |
+
"grad_norm": 1.421875,
|
| 1870 |
+
"learning_rate": 1.5251807805295302e-06,
|
| 1871 |
+
"loss": 0.208,
|
| 1872 |
+
"step": 2660
|
| 1873 |
+
},
|
| 1874 |
+
{
|
| 1875 |
+
"epoch": 0.6675,
|
| 1876 |
+
"grad_norm": 1.125,
|
| 1877 |
+
"learning_rate": 1.5051284217309743e-06,
|
| 1878 |
+
"loss": 0.1951,
|
| 1879 |
+
"step": 2670
|
| 1880 |
+
},
|
| 1881 |
+
{
|
| 1882 |
+
"epoch": 0.67,
|
| 1883 |
+
"grad_norm": 1.0859375,
|
| 1884 |
+
"learning_rate": 1.4851518262549058e-06,
|
| 1885 |
+
"loss": 0.1936,
|
| 1886 |
+
"step": 2680
|
| 1887 |
+
},
|
| 1888 |
+
{
|
| 1889 |
+
"epoch": 0.6725,
|
| 1890 |
+
"grad_norm": 1.25,
|
| 1891 |
+
"learning_rate": 1.465252515396413e-06,
|
| 1892 |
+
"loss": 0.2048,
|
| 1893 |
+
"step": 2690
|
| 1894 |
+
},
|
| 1895 |
+
{
|
| 1896 |
+
"epoch": 0.675,
|
| 1897 |
+
"grad_norm": 1.1328125,
|
| 1898 |
+
"learning_rate": 1.4454320045650606e-06,
|
| 1899 |
+
"loss": 0.1979,
|
| 1900 |
+
"step": 2700
|
| 1901 |
+
},
|
| 1902 |
+
{
|
| 1903 |
+
"epoch": 0.6775,
|
| 1904 |
+
"grad_norm": 1.265625,
|
| 1905 |
+
"learning_rate": 1.4256918031694866e-06,
|
| 1906 |
+
"loss": 0.198,
|
| 1907 |
+
"step": 2710
|
| 1908 |
+
},
|
| 1909 |
+
{
|
| 1910 |
+
"epoch": 0.68,
|
| 1911 |
+
"grad_norm": 1.15625,
|
| 1912 |
+
"learning_rate": 1.4060334145024543e-06,
|
| 1913 |
+
"loss": 0.1902,
|
| 1914 |
+
"step": 2720
|
| 1915 |
+
},
|
| 1916 |
+
{
|
| 1917 |
+
"epoch": 0.6825,
|
| 1918 |
+
"grad_norm": 1.046875,
|
| 1919 |
+
"learning_rate": 1.3864583356263706e-06,
|
| 1920 |
+
"loss": 0.1911,
|
| 1921 |
+
"step": 2730
|
| 1922 |
+
},
|
| 1923 |
+
{
|
| 1924 |
+
"epoch": 0.685,
|
| 1925 |
+
"grad_norm": 1.15625,
|
| 1926 |
+
"learning_rate": 1.366968057259282e-06,
|
| 1927 |
+
"loss": 0.2109,
|
| 1928 |
+
"step": 2740
|
| 1929 |
+
},
|
| 1930 |
+
{
|
| 1931 |
+
"epoch": 0.6875,
|
| 1932 |
+
"grad_norm": 1.140625,
|
| 1933 |
+
"learning_rate": 1.3475640636613447e-06,
|
| 1934 |
+
"loss": 0.2015,
|
| 1935 |
+
"step": 2750
|
| 1936 |
+
},
|
| 1937 |
+
{
|
| 1938 |
+
"epoch": 0.69,
|
| 1939 |
+
"grad_norm": 1.265625,
|
| 1940 |
+
"learning_rate": 1.3282478325217961e-06,
|
| 1941 |
+
"loss": 0.2006,
|
| 1942 |
+
"step": 2760
|
| 1943 |
+
},
|
| 1944 |
+
{
|
| 1945 |
+
"epoch": 0.6925,
|
| 1946 |
+
"grad_norm": 1.046875,
|
| 1947 |
+
"learning_rate": 1.3090208348464244e-06,
|
| 1948 |
+
"loss": 0.1842,
|
| 1949 |
+
"step": 2770
|
| 1950 |
+
},
|
| 1951 |
+
{
|
| 1952 |
+
"epoch": 0.695,
|
| 1953 |
+
"grad_norm": 1.2890625,
|
| 1954 |
+
"learning_rate": 1.289884534845542e-06,
|
| 1955 |
+
"loss": 0.2091,
|
| 1956 |
+
"step": 2780
|
| 1957 |
+
},
|
| 1958 |
+
{
|
| 1959 |
+
"epoch": 0.6975,
|
| 1960 |
+
"grad_norm": 0.953125,
|
| 1961 |
+
"learning_rate": 1.2708403898224839e-06,
|
| 1962 |
+
"loss": 0.185,
|
| 1963 |
+
"step": 2790
|
| 1964 |
+
},
|
| 1965 |
+
{
|
| 1966 |
+
"epoch": 0.7,
|
| 1967 |
+
"grad_norm": 1.0625,
|
| 1968 |
+
"learning_rate": 1.2518898500626259e-06,
|
| 1969 |
+
"loss": 0.1875,
|
| 1970 |
+
"step": 2800
|
| 1971 |
+
},
|
| 1972 |
+
{
|
| 1973 |
+
"epoch": 0.7025,
|
| 1974 |
+
"grad_norm": 1.140625,
|
| 1975 |
+
"learning_rate": 1.2330343587229397e-06,
|
| 1976 |
+
"loss": 0.1926,
|
| 1977 |
+
"step": 2810
|
| 1978 |
+
},
|
| 1979 |
+
{
|
| 1980 |
+
"epoch": 0.705,
|
| 1981 |
+
"grad_norm": 1.234375,
|
| 1982 |
+
"learning_rate": 1.2142753517220945e-06,
|
| 1983 |
+
"loss": 0.1988,
|
| 1984 |
+
"step": 2820
|
| 1985 |
+
},
|
| 1986 |
+
{
|
| 1987 |
+
"epoch": 0.7075,
|
| 1988 |
+
"grad_norm": 1.046875,
|
| 1989 |
+
"learning_rate": 1.1956142576311011e-06,
|
| 1990 |
+
"loss": 0.1988,
|
| 1991 |
+
"step": 2830
|
| 1992 |
+
},
|
| 1993 |
+
{
|
| 1994 |
+
"epoch": 0.71,
|
| 1995 |
+
"grad_norm": 1.1484375,
|
| 1996 |
+
"learning_rate": 1.177052497564524e-06,
|
| 1997 |
+
"loss": 0.1817,
|
| 1998 |
+
"step": 2840
|
| 1999 |
+
},
|
| 2000 |
+
{
|
| 2001 |
+
"epoch": 0.7125,
|
| 2002 |
+
"grad_norm": 1.2109375,
|
| 2003 |
+
"learning_rate": 1.1585914850722565e-06,
|
| 2004 |
+
"loss": 0.1887,
|
| 2005 |
+
"step": 2850
|
| 2006 |
+
},
|
| 2007 |
+
{
|
| 2008 |
+
"epoch": 0.715,
|
| 2009 |
+
"grad_norm": 1.140625,
|
| 2010 |
+
"learning_rate": 1.1402326260318752e-06,
|
| 2011 |
+
"loss": 0.1821,
|
| 2012 |
+
"step": 2860
|
| 2013 |
+
},
|
| 2014 |
+
{
|
| 2015 |
+
"epoch": 0.7175,
|
| 2016 |
+
"grad_norm": 1.234375,
|
| 2017 |
+
"learning_rate": 1.121977318541575e-06,
|
| 2018 |
+
"loss": 0.1948,
|
| 2019 |
+
"step": 2870
|
| 2020 |
+
},
|
| 2021 |
+
{
|
| 2022 |
+
"epoch": 0.72,
|
| 2023 |
+
"grad_norm": 1.2265625,
|
| 2024 |
+
"learning_rate": 1.1038269528136989e-06,
|
| 2025 |
+
"loss": 0.2054,
|
| 2026 |
+
"step": 2880
|
| 2027 |
+
},
|
| 2028 |
+
{
|
| 2029 |
+
"epoch": 0.7225,
|
| 2030 |
+
"grad_norm": 1.0859375,
|
| 2031 |
+
"learning_rate": 1.0857829110688695e-06,
|
| 2032 |
+
"loss": 0.1956,
|
| 2033 |
+
"step": 2890
|
| 2034 |
+
},
|
| 2035 |
+
{
|
| 2036 |
+
"epoch": 0.725,
|
| 2037 |
+
"grad_norm": 1.09375,
|
| 2038 |
+
"learning_rate": 1.0678465674307273e-06,
|
| 2039 |
+
"loss": 0.2033,
|
| 2040 |
+
"step": 2900
|
| 2041 |
+
},
|
| 2042 |
+
{
|
| 2043 |
+
"epoch": 0.7275,
|
| 2044 |
+
"grad_norm": 1.046875,
|
| 2045 |
+
"learning_rate": 1.0500192878212826e-06,
|
| 2046 |
+
"loss": 0.1986,
|
| 2047 |
+
"step": 2910
|
| 2048 |
+
},
|
| 2049 |
+
{
|
| 2050 |
+
"epoch": 0.73,
|
| 2051 |
+
"grad_norm": 1.1328125,
|
| 2052 |
+
"learning_rate": 1.032302429856899e-06,
|
| 2053 |
+
"loss": 0.1979,
|
| 2054 |
+
"step": 2920
|
| 2055 |
+
},
|
| 2056 |
+
{
|
| 2057 |
+
"epoch": 0.7325,
|
| 2058 |
+
"grad_norm": 1.28125,
|
| 2059 |
+
"learning_rate": 1.014697342744904e-06,
|
| 2060 |
+
"loss": 0.1994,
|
| 2061 |
+
"step": 2930
|
| 2062 |
+
},
|
| 2063 |
+
{
|
| 2064 |
+
"epoch": 0.735,
|
| 2065 |
+
"grad_norm": 1.0859375,
|
| 2066 |
+
"learning_rate": 9.97205367180842e-07,
|
| 2067 |
+
"loss": 0.1921,
|
| 2068 |
+
"step": 2940
|
| 2069 |
+
},
|
| 2070 |
+
{
|
| 2071 |
+
"epoch": 0.7375,
|
| 2072 |
+
"grad_norm": 1.1328125,
|
| 2073 |
+
"learning_rate": 9.798278352463752e-07,
|
| 2074 |
+
"loss": 0.1999,
|
| 2075 |
+
"step": 2950
|
| 2076 |
+
},
|
| 2077 |
+
{
|
| 2078 |
+
"epoch": 0.74,
|
| 2079 |
+
"grad_norm": 1.1015625,
|
| 2080 |
+
"learning_rate": 9.625660703078394e-07,
|
| 2081 |
+
"loss": 0.2021,
|
| 2082 |
+
"step": 2960
|
| 2083 |
+
},
|
| 2084 |
+
{
|
| 2085 |
+
"epoch": 0.7425,
|
| 2086 |
+
"grad_norm": 1.296875,
|
| 2087 |
+
"learning_rate": 9.45421386915468e-07,
|
| 2088 |
+
"loss": 0.2027,
|
| 2089 |
+
"step": 2970
|
| 2090 |
+
},
|
| 2091 |
+
{
|
| 2092 |
+
"epoch": 0.745,
|
| 2093 |
+
"grad_norm": 1.34375,
|
| 2094 |
+
"learning_rate": 9.283950907032788e-07,
|
| 2095 |
+
"loss": 0.1898,
|
| 2096 |
+
"step": 2980
|
| 2097 |
+
},
|
| 2098 |
+
{
|
| 2099 |
+
"epoch": 0.7475,
|
| 2100 |
+
"grad_norm": 1.1484375,
|
| 2101 |
+
"learning_rate": 9.114884782896482e-07,
|
| 2102 |
+
"loss": 0.1943,
|
| 2103 |
+
"step": 2990
|
| 2104 |
+
},
|
| 2105 |
+
{
|
| 2106 |
+
"epoch": 0.75,
|
| 2107 |
+
"grad_norm": 1.3515625,
|
| 2108 |
+
"learning_rate": 8.947028371785677e-07,
|
| 2109 |
+
"loss": 0.2189,
|
| 2110 |
+
"step": 3000
|
| 2111 |
+
},
|
| 2112 |
+
{
|
| 2113 |
+
"epoch": 0.7525,
|
| 2114 |
+
"grad_norm": 1.0703125,
|
| 2115 |
+
"learning_rate": 8.780394456615974e-07,
|
| 2116 |
+
"loss": 0.1889,
|
| 2117 |
+
"step": 3010
|
| 2118 |
+
},
|
| 2119 |
+
{
|
| 2120 |
+
"epoch": 0.755,
|
| 2121 |
+
"grad_norm": 1.3359375,
|
| 2122 |
+
"learning_rate": 8.614995727205155e-07,
|
| 2123 |
+
"loss": 0.2013,
|
| 2124 |
+
"step": 3020
|
| 2125 |
+
},
|
| 2126 |
+
{
|
| 2127 |
+
"epoch": 0.7575,
|
| 2128 |
+
"grad_norm": 1.0703125,
|
| 2129 |
+
"learning_rate": 8.450844779306827e-07,
|
| 2130 |
+
"loss": 0.1867,
|
| 2131 |
+
"step": 3030
|
| 2132 |
+
},
|
| 2133 |
+
{
|
| 2134 |
+
"epoch": 0.76,
|
| 2135 |
+
"grad_norm": 1.0234375,
|
| 2136 |
+
"learning_rate": 8.28795411365122e-07,
|
| 2137 |
+
"loss": 0.2004,
|
| 2138 |
+
"step": 3040
|
| 2139 |
+
},
|
| 2140 |
+
{
|
| 2141 |
+
"epoch": 0.7625,
|
| 2142 |
+
"grad_norm": 1.484375,
|
| 2143 |
+
"learning_rate": 8.126336134993176e-07,
|
| 2144 |
+
"loss": 0.1959,
|
| 2145 |
+
"step": 3050
|
| 2146 |
+
},
|
| 2147 |
+
{
|
| 2148 |
+
"epoch": 0.765,
|
| 2149 |
+
"grad_norm": 1.1484375,
|
| 2150 |
+
"learning_rate": 7.966003151167498e-07,
|
| 2151 |
+
"loss": 0.1855,
|
| 2152 |
+
"step": 3060
|
| 2153 |
+
},
|
| 2154 |
+
{
|
| 2155 |
+
"epoch": 0.7675,
|
| 2156 |
+
"grad_norm": 1.125,
|
| 2157 |
+
"learning_rate": 7.806967372151661e-07,
|
| 2158 |
+
"loss": 0.1901,
|
| 2159 |
+
"step": 3070
|
| 2160 |
+
},
|
| 2161 |
+
{
|
| 2162 |
+
"epoch": 0.77,
|
| 2163 |
+
"grad_norm": 1.609375,
|
| 2164 |
+
"learning_rate": 7.649240909135966e-07,
|
| 2165 |
+
"loss": 0.2125,
|
| 2166 |
+
"step": 3080
|
| 2167 |
+
},
|
| 2168 |
+
{
|
| 2169 |
+
"epoch": 0.7725,
|
| 2170 |
+
"grad_norm": 1.171875,
|
| 2171 |
+
"learning_rate": 7.492835773601234e-07,
|
| 2172 |
+
"loss": 0.1916,
|
| 2173 |
+
"step": 3090
|
| 2174 |
+
},
|
| 2175 |
+
{
|
| 2176 |
+
"epoch": 0.775,
|
| 2177 |
+
"grad_norm": 1.140625,
|
| 2178 |
+
"learning_rate": 7.337763876404078e-07,
|
| 2179 |
+
"loss": 0.1844,
|
| 2180 |
+
"step": 3100
|
| 2181 |
+
},
|
| 2182 |
+
{
|
| 2183 |
+
"epoch": 0.7775,
|
| 2184 |
+
"grad_norm": 1.0390625,
|
| 2185 |
+
"learning_rate": 7.184037026869867e-07,
|
| 2186 |
+
"loss": 0.1908,
|
| 2187 |
+
"step": 3110
|
| 2188 |
+
},
|
| 2189 |
+
{
|
| 2190 |
+
"epoch": 0.78,
|
| 2191 |
+
"grad_norm": 1.1015625,
|
| 2192 |
+
"learning_rate": 7.031666931893361e-07,
|
| 2193 |
+
"loss": 0.1858,
|
| 2194 |
+
"step": 3120
|
| 2195 |
+
},
|
| 2196 |
+
{
|
| 2197 |
+
"epoch": 0.7825,
|
| 2198 |
+
"grad_norm": 1.171875,
|
| 2199 |
+
"learning_rate": 6.880665195047226e-07,
|
| 2200 |
+
"loss": 0.1976,
|
| 2201 |
+
"step": 3130
|
| 2202 |
+
},
|
| 2203 |
+
{
|
| 2204 |
+
"epoch": 0.785,
|
| 2205 |
+
"grad_norm": 1.1015625,
|
| 2206 |
+
"learning_rate": 6.731043315698346e-07,
|
| 2207 |
+
"loss": 0.1901,
|
| 2208 |
+
"step": 3140
|
| 2209 |
+
},
|
| 2210 |
+
{
|
| 2211 |
+
"epoch": 0.7875,
|
| 2212 |
+
"grad_norm": 1.15625,
|
| 2213 |
+
"learning_rate": 6.58281268813212e-07,
|
| 2214 |
+
"loss": 0.2021,
|
| 2215 |
+
"step": 3150
|
| 2216 |
+
},
|
| 2217 |
+
{
|
| 2218 |
+
"epoch": 0.79,
|
| 2219 |
+
"grad_norm": 1.4296875,
|
| 2220 |
+
"learning_rate": 6.435984600684731e-07,
|
| 2221 |
+
"loss": 0.2088,
|
| 2222 |
+
"step": 3160
|
| 2223 |
+
},
|
| 2224 |
+
{
|
| 2225 |
+
"epoch": 0.7925,
|
| 2226 |
+
"grad_norm": 1.0390625,
|
| 2227 |
+
"learning_rate": 6.290570234883506e-07,
|
| 2228 |
+
"loss": 0.2004,
|
| 2229 |
+
"step": 3170
|
| 2230 |
+
},
|
| 2231 |
+
{
|
| 2232 |
+
"epoch": 0.795,
|
| 2233 |
+
"grad_norm": 4.46875,
|
| 2234 |
+
"learning_rate": 6.146580664595391e-07,
|
| 2235 |
+
"loss": 0.1964,
|
| 2236 |
+
"step": 3180
|
| 2237 |
+
},
|
| 2238 |
+
{
|
| 2239 |
+
"epoch": 0.7975,
|
| 2240 |
+
"grad_norm": 1.15625,
|
| 2241 |
+
"learning_rate": 6.004026855183656e-07,
|
| 2242 |
+
"loss": 0.1996,
|
| 2243 |
+
"step": 3190
|
| 2244 |
+
},
|
| 2245 |
+
{
|
| 2246 |
+
"epoch": 0.8,
|
| 2247 |
+
"grad_norm": 1.0390625,
|
| 2248 |
+
"learning_rate": 5.862919662672801e-07,
|
| 2249 |
+
"loss": 0.1919,
|
| 2250 |
+
"step": 3200
|
| 2251 |
+
},
|
| 2252 |
+
{
|
| 2253 |
+
"epoch": 0.8025,
|
| 2254 |
+
"grad_norm": 1.1796875,
|
| 2255 |
+
"learning_rate": 5.723269832921849e-07,
|
| 2256 |
+
"loss": 0.202,
|
| 2257 |
+
"step": 3210
|
| 2258 |
+
},
|
| 2259 |
+
{
|
| 2260 |
+
"epoch": 0.805,
|
| 2261 |
+
"grad_norm": 1.1015625,
|
| 2262 |
+
"learning_rate": 5.585088000806016e-07,
|
| 2263 |
+
"loss": 0.2024,
|
| 2264 |
+
"step": 3220
|
| 2265 |
+
},
|
| 2266 |
+
{
|
| 2267 |
+
"epoch": 0.8075,
|
| 2268 |
+
"grad_norm": 1.2734375,
|
| 2269 |
+
"learning_rate": 5.448384689406804e-07,
|
| 2270 |
+
"loss": 0.1962,
|
| 2271 |
+
"step": 3230
|
| 2272 |
+
},
|
| 2273 |
+
{
|
| 2274 |
+
"epoch": 0.81,
|
| 2275 |
+
"grad_norm": 1.2890625,
|
| 2276 |
+
"learning_rate": 5.313170309210655e-07,
|
| 2277 |
+
"loss": 0.1879,
|
| 2278 |
+
"step": 3240
|
| 2279 |
+
},
|
| 2280 |
+
{
|
| 2281 |
+
"epoch": 0.8125,
|
| 2282 |
+
"grad_norm": 0.99609375,
|
| 2283 |
+
"learning_rate": 5.179455157316124e-07,
|
| 2284 |
+
"loss": 0.1987,
|
| 2285 |
+
"step": 3250
|
| 2286 |
+
},
|
| 2287 |
+
{
|
| 2288 |
+
"epoch": 0.815,
|
| 2289 |
+
"grad_norm": 1.15625,
|
| 2290 |
+
"learning_rate": 5.047249416649713e-07,
|
| 2291 |
+
"loss": 0.2078,
|
| 2292 |
+
"step": 3260
|
| 2293 |
+
},
|
| 2294 |
+
{
|
| 2295 |
+
"epoch": 0.8175,
|
| 2296 |
+
"grad_norm": 14.25,
|
| 2297 |
+
"learning_rate": 4.916563155190446e-07,
|
| 2298 |
+
"loss": 0.2002,
|
| 2299 |
+
"step": 3270
|
| 2300 |
+
},
|
| 2301 |
+
{
|
| 2302 |
+
"epoch": 0.82,
|
| 2303 |
+
"grad_norm": 1.09375,
|
| 2304 |
+
"learning_rate": 4.787406325203101e-07,
|
| 2305 |
+
"loss": 0.1948,
|
| 2306 |
+
"step": 3280
|
| 2307 |
+
},
|
| 2308 |
+
{
|
| 2309 |
+
"epoch": 0.8225,
|
| 2310 |
+
"grad_norm": 1.0859375,
|
| 2311 |
+
"learning_rate": 4.6597887624803273e-07,
|
| 2312 |
+
"loss": 0.1981,
|
| 2313 |
+
"step": 3290
|
| 2314 |
+
},
|
| 2315 |
+
{
|
| 2316 |
+
"epoch": 0.825,
|
| 2317 |
+
"grad_norm": 1.1796875,
|
| 2318 |
+
"learning_rate": 4.533720185593621e-07,
|
| 2319 |
+
"loss": 0.1965,
|
| 2320 |
+
"step": 3300
|
| 2321 |
+
},
|
| 2322 |
+
{
|
| 2323 |
+
"epoch": 0.8275,
|
| 2324 |
+
"grad_norm": 1.1875,
|
| 2325 |
+
"learning_rate": 4.4092101951532076e-07,
|
| 2326 |
+
"loss": 0.1926,
|
| 2327 |
+
"step": 3310
|
| 2328 |
+
},
|
| 2329 |
+
{
|
| 2330 |
+
"epoch": 0.83,
|
| 2331 |
+
"grad_norm": 1.2578125,
|
| 2332 |
+
"learning_rate": 4.2862682730769157e-07,
|
| 2333 |
+
"loss": 0.1952,
|
| 2334 |
+
"step": 3320
|
| 2335 |
+
},
|
| 2336 |
+
{
|
| 2337 |
+
"epoch": 0.8325,
|
| 2338 |
+
"grad_norm": 1.1484375,
|
| 2339 |
+
"learning_rate": 4.164903781868096e-07,
|
| 2340 |
+
"loss": 0.1853,
|
| 2341 |
+
"step": 3330
|
| 2342 |
+
},
|
| 2343 |
+
{
|
| 2344 |
+
"epoch": 0.835,
|
| 2345 |
+
"grad_norm": 1.6015625,
|
| 2346 |
+
"learning_rate": 4.045125963902641e-07,
|
| 2347 |
+
"loss": 0.2008,
|
| 2348 |
+
"step": 3340
|
| 2349 |
+
},
|
| 2350 |
+
{
|
| 2351 |
+
"epoch": 0.8375,
|
| 2352 |
+
"grad_norm": 1.1015625,
|
| 2353 |
+
"learning_rate": 3.9269439407251365e-07,
|
| 2354 |
+
"loss": 0.2064,
|
| 2355 |
+
"step": 3350
|
| 2356 |
+
},
|
| 2357 |
+
{
|
| 2358 |
+
"epoch": 0.84,
|
| 2359 |
+
"grad_norm": 1.1640625,
|
| 2360 |
+
"learning_rate": 3.810366712354199e-07,
|
| 2361 |
+
"loss": 0.1833,
|
| 2362 |
+
"step": 3360
|
| 2363 |
+
},
|
| 2364 |
+
{
|
| 2365 |
+
"epoch": 0.8425,
|
| 2366 |
+
"grad_norm": 1.0546875,
|
| 2367 |
+
"learning_rate": 3.6954031565971187e-07,
|
| 2368 |
+
"loss": 0.1889,
|
| 2369 |
+
"step": 3370
|
| 2370 |
+
},
|
| 2371 |
+
{
|
| 2372 |
+
"epoch": 0.845,
|
| 2373 |
+
"grad_norm": 0.984375,
|
| 2374 |
+
"learning_rate": 3.5820620283737615e-07,
|
| 2375 |
+
"loss": 0.1785,
|
| 2376 |
+
"step": 3380
|
| 2377 |
+
},
|
| 2378 |
+
{
|
| 2379 |
+
"epoch": 0.8475,
|
| 2380 |
+
"grad_norm": 1.46875,
|
| 2381 |
+
"learning_rate": 3.4703519590498615e-07,
|
| 2382 |
+
"loss": 0.2022,
|
| 2383 |
+
"step": 3390
|
| 2384 |
+
},
|
| 2385 |
+
{
|
| 2386 |
+
"epoch": 0.85,
|
| 2387 |
+
"grad_norm": 1.140625,
|
| 2388 |
+
"learning_rate": 3.360281455779704e-07,
|
| 2389 |
+
"loss": 0.1942,
|
| 2390 |
+
"step": 3400
|
| 2391 |
+
},
|
| 2392 |
+
{
|
| 2393 |
+
"epoch": 0.8525,
|
| 2394 |
+
"grad_norm": 1.1640625,
|
| 2395 |
+
"learning_rate": 3.2518589008582597e-07,
|
| 2396 |
+
"loss": 0.1933,
|
| 2397 |
+
"step": 3410
|
| 2398 |
+
},
|
| 2399 |
+
{
|
| 2400 |
+
"epoch": 0.855,
|
| 2401 |
+
"grad_norm": 1.1875,
|
| 2402 |
+
"learning_rate": 3.1450925510828705e-07,
|
| 2403 |
+
"loss": 0.1902,
|
| 2404 |
+
"step": 3420
|
| 2405 |
+
},
|
| 2406 |
+
{
|
| 2407 |
+
"epoch": 0.8575,
|
| 2408 |
+
"grad_norm": 1.1796875,
|
| 2409 |
+
"learning_rate": 3.039990537124432e-07,
|
| 2410 |
+
"loss": 0.2161,
|
| 2411 |
+
"step": 3430
|
| 2412 |
+
},
|
| 2413 |
+
{
|
| 2414 |
+
"epoch": 0.86,
|
| 2415 |
+
"grad_norm": 1.078125,
|
| 2416 |
+
"learning_rate": 2.936560862908225e-07,
|
| 2417 |
+
"loss": 0.178,
|
| 2418 |
+
"step": 3440
|
| 2419 |
+
},
|
| 2420 |
+
{
|
| 2421 |
+
"epoch": 0.8625,
|
| 2422 |
+
"grad_norm": 1.234375,
|
| 2423 |
+
"learning_rate": 2.8348114050043813e-07,
|
| 2424 |
+
"loss": 0.2021,
|
| 2425 |
+
"step": 3450
|
| 2426 |
+
},
|
| 2427 |
+
{
|
| 2428 |
+
"epoch": 0.865,
|
| 2429 |
+
"grad_norm": 1.578125,
|
| 2430 |
+
"learning_rate": 2.7347499120280677e-07,
|
| 2431 |
+
"loss": 0.2043,
|
| 2432 |
+
"step": 3460
|
| 2433 |
+
},
|
| 2434 |
+
{
|
| 2435 |
+
"epoch": 0.8675,
|
| 2436 |
+
"grad_norm": 1.328125,
|
| 2437 |
+
"learning_rate": 2.636384004049375e-07,
|
| 2438 |
+
"loss": 0.1921,
|
| 2439 |
+
"step": 3470
|
| 2440 |
+
},
|
| 2441 |
+
{
|
| 2442 |
+
"epoch": 0.87,
|
| 2443 |
+
"grad_norm": 1.125,
|
| 2444 |
+
"learning_rate": 2.5397211720130267e-07,
|
| 2445 |
+
"loss": 0.19,
|
| 2446 |
+
"step": 3480
|
| 2447 |
+
},
|
| 2448 |
+
{
|
| 2449 |
+
"epoch": 0.8725,
|
| 2450 |
+
"grad_norm": 1.171875,
|
| 2451 |
+
"learning_rate": 2.4447687771679414e-07,
|
| 2452 |
+
"loss": 0.1914,
|
| 2453 |
+
"step": 3490
|
| 2454 |
+
},
|
| 2455 |
+
{
|
| 2456 |
+
"epoch": 0.875,
|
| 2457 |
+
"grad_norm": 1.09375,
|
| 2458 |
+
"learning_rate": 2.3515340505066043e-07,
|
| 2459 |
+
"loss": 0.1919,
|
| 2460 |
+
"step": 3500
|
| 2461 |
+
},
|
| 2462 |
+
{
|
| 2463 |
+
"epoch": 0.8775,
|
| 2464 |
+
"grad_norm": 1.0859375,
|
| 2465 |
+
"learning_rate": 2.260024092214419e-07,
|
| 2466 |
+
"loss": 0.2006,
|
| 2467 |
+
"step": 3510
|
| 2468 |
+
},
|
| 2469 |
+
{
|
| 2470 |
+
"epoch": 0.88,
|
| 2471 |
+
"grad_norm": 1.1640625,
|
| 2472 |
+
"learning_rate": 2.170245871129012e-07,
|
| 2473 |
+
"loss": 0.1982,
|
| 2474 |
+
"step": 3520
|
| 2475 |
+
},
|
| 2476 |
+
{
|
| 2477 |
+
"epoch": 0.8825,
|
| 2478 |
+
"grad_norm": 1.9453125,
|
| 2479 |
+
"learning_rate": 2.0822062242095015e-07,
|
| 2480 |
+
"loss": 0.2113,
|
| 2481 |
+
"step": 3530
|
| 2482 |
+
},
|
| 2483 |
+
{
|
| 2484 |
+
"epoch": 0.885,
|
| 2485 |
+
"grad_norm": 1.1171875,
|
| 2486 |
+
"learning_rate": 1.995911856015867e-07,
|
| 2487 |
+
"loss": 0.1969,
|
| 2488 |
+
"step": 3540
|
| 2489 |
+
},
|
| 2490 |
+
{
|
| 2491 |
+
"epoch": 0.8875,
|
| 2492 |
+
"grad_norm": 1.1796875,
|
| 2493 |
+
"learning_rate": 1.9113693381983405e-07,
|
| 2494 |
+
"loss": 0.1974,
|
| 2495 |
+
"step": 3550
|
| 2496 |
+
},
|
| 2497 |
+
{
|
| 2498 |
+
"epoch": 0.89,
|
| 2499 |
+
"grad_norm": 1.203125,
|
| 2500 |
+
"learning_rate": 1.8285851089969802e-07,
|
| 2501 |
+
"loss": 0.1987,
|
| 2502 |
+
"step": 3560
|
| 2503 |
+
},
|
| 2504 |
+
{
|
| 2505 |
+
"epoch": 0.8925,
|
| 2506 |
+
"grad_norm": 1.3046875,
|
| 2507 |
+
"learning_rate": 1.7475654727513502e-07,
|
| 2508 |
+
"loss": 0.201,
|
| 2509 |
+
"step": 3570
|
| 2510 |
+
},
|
| 2511 |
+
{
|
| 2512 |
+
"epoch": 0.895,
|
| 2513 |
+
"grad_norm": 1.21875,
|
| 2514 |
+
"learning_rate": 1.668316599420433e-07,
|
| 2515 |
+
"loss": 0.185,
|
| 2516 |
+
"step": 3580
|
| 2517 |
+
},
|
| 2518 |
+
{
|
| 2519 |
+
"epoch": 0.8975,
|
| 2520 |
+
"grad_norm": 1.0703125,
|
| 2521 |
+
"learning_rate": 1.5908445241127528e-07,
|
| 2522 |
+
"loss": 0.1872,
|
| 2523 |
+
"step": 3590
|
| 2524 |
+
},
|
| 2525 |
+
{
|
| 2526 |
+
"epoch": 0.9,
|
| 2527 |
+
"grad_norm": 1.1953125,
|
| 2528 |
+
"learning_rate": 1.5151551466267956e-07,
|
| 2529 |
+
"loss": 0.1985,
|
| 2530 |
+
"step": 3600
|
| 2531 |
+
},
|
| 2532 |
+
{
|
| 2533 |
+
"epoch": 0.9025,
|
| 2534 |
+
"grad_norm": 1.1796875,
|
| 2535 |
+
"learning_rate": 1.441254231001696e-07,
|
| 2536 |
+
"loss": 0.2036,
|
| 2537 |
+
"step": 3610
|
| 2538 |
+
},
|
| 2539 |
+
{
|
| 2540 |
+
"epoch": 0.905,
|
| 2541 |
+
"grad_norm": 1.125,
|
| 2542 |
+
"learning_rate": 1.3691474050782972e-07,
|
| 2543 |
+
"loss": 0.1902,
|
| 2544 |
+
"step": 3620
|
| 2545 |
+
},
|
| 2546 |
+
{
|
| 2547 |
+
"epoch": 0.9075,
|
| 2548 |
+
"grad_norm": 1.078125,
|
| 2549 |
+
"learning_rate": 1.2988401600705635e-07,
|
| 2550 |
+
"loss": 0.1921,
|
| 2551 |
+
"step": 3630
|
| 2552 |
+
},
|
| 2553 |
+
{
|
| 2554 |
+
"epoch": 0.91,
|
| 2555 |
+
"grad_norm": 1.1640625,
|
| 2556 |
+
"learning_rate": 1.2303378501474174e-07,
|
| 2557 |
+
"loss": 0.1752,
|
| 2558 |
+
"step": 3640
|
| 2559 |
+
},
|
| 2560 |
+
{
|
| 2561 |
+
"epoch": 0.9125,
|
| 2562 |
+
"grad_norm": 1.1953125,
|
| 2563 |
+
"learning_rate": 1.16364569202497e-07,
|
| 2564 |
+
"loss": 0.1893,
|
| 2565 |
+
"step": 3650
|
| 2566 |
+
},
|
| 2567 |
+
{
|
| 2568 |
+
"epoch": 0.915,
|
| 2569 |
+
"grad_norm": 1.1953125,
|
| 2570 |
+
"learning_rate": 1.0987687645692746e-07,
|
| 2571 |
+
"loss": 0.2005,
|
| 2572 |
+
"step": 3660
|
| 2573 |
+
},
|
| 2574 |
+
{
|
| 2575 |
+
"epoch": 0.9175,
|
| 2576 |
+
"grad_norm": 1.4375,
|
| 2577 |
+
"learning_rate": 1.035712008409534e-07,
|
| 2578 |
+
"loss": 0.1948,
|
| 2579 |
+
"step": 3670
|
| 2580 |
+
},
|
| 2581 |
+
{
|
| 2582 |
+
"epoch": 0.92,
|
| 2583 |
+
"grad_norm": 1.125,
|
| 2584 |
+
"learning_rate": 9.744802255618662e-08,
|
| 2585 |
+
"loss": 0.1903,
|
| 2586 |
+
"step": 3680
|
| 2587 |
+
},
|
| 2588 |
+
{
|
| 2589 |
+
"epoch": 0.9225,
|
| 2590 |
+
"grad_norm": 1.109375,
|
| 2591 |
+
"learning_rate": 9.150780790636054e-08,
|
| 2592 |
+
"loss": 0.2147,
|
| 2593 |
+
"step": 3690
|
| 2594 |
+
},
|
| 2595 |
+
{
|
| 2596 |
+
"epoch": 0.925,
|
| 2597 |
+
"grad_norm": 1.2734375,
|
| 2598 |
+
"learning_rate": 8.575100926181884e-08,
|
| 2599 |
+
"loss": 0.19,
|
| 2600 |
+
"step": 3700
|
| 2601 |
+
},
|
| 2602 |
+
{
|
| 2603 |
+
"epoch": 0.9275,
|
| 2604 |
+
"grad_norm": 1.09375,
|
| 2605 |
+
"learning_rate": 8.017806502506692e-08,
|
| 2606 |
+
"loss": 0.2004,
|
| 2607 |
+
"step": 3710
|
| 2608 |
+
},
|
| 2609 |
+
{
|
| 2610 |
+
"epoch": 0.93,
|
| 2611 |
+
"grad_norm": 1.2421875,
|
| 2612 |
+
"learning_rate": 7.478939959738502e-08,
|
| 2613 |
+
"loss": 0.2059,
|
| 2614 |
+
"step": 3720
|
| 2615 |
+
},
|
| 2616 |
+
{
|
| 2617 |
+
"epoch": 0.9325,
|
| 2618 |
+
"grad_norm": 1.234375,
|
| 2619 |
+
"learning_rate": 6.958542334650847e-08,
|
| 2620 |
+
"loss": 0.1908,
|
| 2621 |
+
"step": 3730
|
| 2622 |
+
},
|
| 2623 |
+
{
|
| 2624 |
+
"epoch": 0.935,
|
| 2625 |
+
"grad_norm": 1.6328125,
|
| 2626 |
+
"learning_rate": 6.456653257537665e-08,
|
| 2627 |
+
"loss": 0.1979,
|
| 2628 |
+
"step": 3740
|
| 2629 |
+
},
|
| 2630 |
+
{
|
| 2631 |
+
"epoch": 0.9375,
|
| 2632 |
+
"grad_norm": 1.0859375,
|
| 2633 |
+
"learning_rate": 5.973310949195343e-08,
|
| 2634 |
+
"loss": 0.1934,
|
| 2635 |
+
"step": 3750
|
| 2636 |
+
},
|
| 2637 |
+
{
|
| 2638 |
+
"epoch": 0.94,
|
| 2639 |
+
"grad_norm": 1.078125,
|
| 2640 |
+
"learning_rate": 5.50855221801197e-08,
|
| 2641 |
+
"loss": 0.1893,
|
| 2642 |
+
"step": 3760
|
| 2643 |
+
},
|
| 2644 |
+
{
|
| 2645 |
+
"epoch": 0.9425,
|
| 2646 |
+
"grad_norm": 1.15625,
|
| 2647 |
+
"learning_rate": 5.062412457164323e-08,
|
| 2648 |
+
"loss": 0.1985,
|
| 2649 |
+
"step": 3770
|
| 2650 |
+
},
|
| 2651 |
+
{
|
| 2652 |
+
"epoch": 0.945,
|
| 2653 |
+
"grad_norm": 1.1953125,
|
| 2654 |
+
"learning_rate": 4.634925641922472e-08,
|
| 2655 |
+
"loss": 0.2008,
|
| 2656 |
+
"step": 3780
|
| 2657 |
+
},
|
| 2658 |
+
{
|
| 2659 |
+
"epoch": 0.9475,
|
| 2660 |
+
"grad_norm": 1.234375,
|
| 2661 |
+
"learning_rate": 4.226124327062514e-08,
|
| 2662 |
+
"loss": 0.2056,
|
| 2663 |
+
"step": 3790
|
| 2664 |
+
},
|
| 2665 |
+
{
|
| 2666 |
+
"epoch": 0.95,
|
| 2667 |
+
"grad_norm": 1.0625,
|
| 2668 |
+
"learning_rate": 3.836039644387307e-08,
|
| 2669 |
+
"loss": 0.2036,
|
| 2670 |
+
"step": 3800
|
| 2671 |
+
},
|
| 2672 |
+
{
|
| 2673 |
+
"epoch": 0.9525,
|
| 2674 |
+
"grad_norm": 16.25,
|
| 2675 |
+
"learning_rate": 3.4647013003556996e-08,
|
| 2676 |
+
"loss": 0.199,
|
| 2677 |
+
"step": 3810
|
| 2678 |
+
},
|
| 2679 |
+
{
|
| 2680 |
+
"epoch": 0.955,
|
| 2681 |
+
"grad_norm": 1.1015625,
|
| 2682 |
+
"learning_rate": 3.112137573820284e-08,
|
| 2683 |
+
"loss": 0.1917,
|
| 2684 |
+
"step": 3820
|
| 2685 |
+
},
|
| 2686 |
+
{
|
| 2687 |
+
"epoch": 0.9575,
|
| 2688 |
+
"grad_norm": 1.0625,
|
| 2689 |
+
"learning_rate": 2.7783753138738713e-08,
|
| 2690 |
+
"loss": 0.1946,
|
| 2691 |
+
"step": 3830
|
| 2692 |
+
},
|
| 2693 |
+
{
|
| 2694 |
+
"epoch": 0.96,
|
| 2695 |
+
"grad_norm": 1.1171875,
|
| 2696 |
+
"learning_rate": 2.463439937804707e-08,
|
| 2697 |
+
"loss": 0.1816,
|
| 2698 |
+
"step": 3840
|
| 2699 |
+
},
|
| 2700 |
+
{
|
| 2701 |
+
"epoch": 0.9625,
|
| 2702 |
+
"grad_norm": 1.2421875,
|
| 2703 |
+
"learning_rate": 2.1673554291610775e-08,
|
| 2704 |
+
"loss": 0.1859,
|
| 2705 |
+
"step": 3850
|
| 2706 |
+
},
|
| 2707 |
+
{
|
| 2708 |
+
"epoch": 0.965,
|
| 2709 |
+
"grad_norm": 1.375,
|
| 2710 |
+
"learning_rate": 1.8901443359245765e-08,
|
| 2711 |
+
"loss": 0.1993,
|
| 2712 |
+
"step": 3860
|
| 2713 |
+
},
|
| 2714 |
+
{
|
| 2715 |
+
"epoch": 0.9675,
|
| 2716 |
+
"grad_norm": 1.328125,
|
| 2717 |
+
"learning_rate": 1.6318277687932816e-08,
|
| 2718 |
+
"loss": 0.2007,
|
| 2719 |
+
"step": 3870
|
| 2720 |
+
},
|
| 2721 |
+
{
|
| 2722 |
+
"epoch": 0.97,
|
| 2723 |
+
"grad_norm": 1.1484375,
|
| 2724 |
+
"learning_rate": 1.3924253995738769e-08,
|
| 2725 |
+
"loss": 0.1925,
|
| 2726 |
+
"step": 3880
|
| 2727 |
+
},
|
| 2728 |
+
{
|
| 2729 |
+
"epoch": 0.9725,
|
| 2730 |
+
"grad_norm": 1.15625,
|
| 2731 |
+
"learning_rate": 1.1719554596836546e-08,
|
| 2732 |
+
"loss": 0.1996,
|
| 2733 |
+
"step": 3890
|
| 2734 |
+
},
|
| 2735 |
+
{
|
| 2736 |
+
"epoch": 0.975,
|
| 2737 |
+
"grad_norm": 1.125,
|
| 2738 |
+
"learning_rate": 9.704347387620994e-09,
|
| 2739 |
+
"loss": 0.1836,
|
| 2740 |
+
"step": 3900
|
| 2741 |
+
},
|
| 2742 |
+
{
|
| 2743 |
+
"epoch": 0.9775,
|
| 2744 |
+
"grad_norm": 1.1640625,
|
| 2745 |
+
"learning_rate": 7.878785833923819e-09,
|
| 2746 |
+
"loss": 0.1928,
|
| 2747 |
+
"step": 3910
|
| 2748 |
+
},
|
| 2749 |
+
{
|
| 2750 |
+
"epoch": 0.98,
|
| 2751 |
+
"grad_norm": 1.46875,
|
| 2752 |
+
"learning_rate": 6.243008959324892e-09,
|
| 2753 |
+
"loss": 0.1967,
|
| 2754 |
+
"step": 3920
|
| 2755 |
+
},
|
| 2756 |
+
{
|
| 2757 |
+
"epoch": 0.9825,
|
| 2758 |
+
"grad_norm": 1.171875,
|
| 2759 |
+
"learning_rate": 4.797141334566268e-09,
|
| 2760 |
+
"loss": 0.1992,
|
| 2761 |
+
"step": 3930
|
| 2762 |
+
},
|
| 2763 |
+
{
|
| 2764 |
+
"epoch": 0.985,
|
| 2765 |
+
"grad_norm": 1.2734375,
|
| 2766 |
+
"learning_rate": 3.5412930680658876e-09,
|
| 2767 |
+
"loss": 0.1993,
|
| 2768 |
+
"step": 3940
|
| 2769 |
+
},
|
| 2770 |
+
{
|
| 2771 |
+
"epoch": 0.9875,
|
| 2772 |
+
"grad_norm": 1.125,
|
| 2773 |
+
"learning_rate": 2.475559797531224e-09,
|
| 2774 |
+
"loss": 0.1989,
|
| 2775 |
+
"step": 3950
|
| 2776 |
+
},
|
| 2777 |
+
{
|
| 2778 |
+
"epoch": 0.99,
|
| 2779 |
+
"grad_norm": 1.09375,
|
| 2780 |
+
"learning_rate": 1.6000226826770604e-09,
|
| 2781 |
+
"loss": 0.1826,
|
| 2782 |
+
"step": 3960
|
| 2783 |
+
},
|
| 2784 |
+
{
|
| 2785 |
+
"epoch": 0.9925,
|
| 2786 |
+
"grad_norm": 1.171875,
|
| 2787 |
+
"learning_rate": 9.147483990443184e-10,
|
| 2788 |
+
"loss": 0.1969,
|
| 2789 |
+
"step": 3970
|
| 2790 |
+
},
|
| 2791 |
+
{
|
| 2792 |
+
"epoch": 0.995,
|
| 2793 |
+
"grad_norm": 1.21875,
|
| 2794 |
+
"learning_rate": 4.197891329230097e-10,
|
| 2795 |
+
"loss": 0.2085,
|
| 2796 |
+
"step": 3980
|
| 2797 |
+
},
|
| 2798 |
+
{
|
| 2799 |
+
"epoch": 0.9975,
|
| 2800 |
+
"grad_norm": 1.2890625,
|
| 2801 |
+
"learning_rate": 1.1518257737763716e-10,
|
| 2802 |
+
"loss": 0.1928,
|
| 2803 |
+
"step": 3990
|
| 2804 |
+
},
|
| 2805 |
+
{
|
| 2806 |
+
"epoch": 1.0,
|
| 2807 |
+
"grad_norm": 1.125,
|
| 2808 |
+
"learning_rate": 9.51929376435956e-13,
|
| 2809 |
+
"loss": 0.1986,
|
| 2810 |
+
"step": 4000
|
| 2811 |
+
}
|
| 2812 |
+
],
|
| 2813 |
+
"logging_steps": 10,
|
| 2814 |
+
"max_steps": 4000,
|
| 2815 |
+
"num_input_tokens_seen": 0,
|
| 2816 |
+
"num_train_epochs": 9223372036854775807,
|
| 2817 |
+
"save_steps": 2000,
|
| 2818 |
+
"stateful_callbacks": {
|
| 2819 |
+
"TrainerControl": {
|
| 2820 |
+
"args": {
|
| 2821 |
+
"should_epoch_stop": false,
|
| 2822 |
+
"should_evaluate": false,
|
| 2823 |
+
"should_log": false,
|
| 2824 |
+
"should_save": true,
|
| 2825 |
+
"should_training_stop": true
|
| 2826 |
+
},
|
| 2827 |
+
"attributes": {}
|
| 2828 |
+
}
|
| 2829 |
+
},
|
| 2830 |
+
"total_flos": 3.82656621576192e+17,
|
| 2831 |
+
"train_batch_size": 32,
|
| 2832 |
+
"trial_name": null,
|
| 2833 |
+
"trial_params": null
|
| 2834 |
+
}
|
checkpoint-4000/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1b39a0514ca81f3fe43bcb544ca347a175b7f15b186cbb3ab69a755fcc316a88
|
| 3 |
+
size 5432
|