See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: EleutherAI/gpt-neo-125m
bf16: true
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - f1bfae7be46056e8_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/f1bfae7be46056e8_train_data.json
  type:
    field_input: intent
    field_instruction: instruction
    field_output: response_8b_instruct
    format: '{instruction} {input}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
deepspeed: null
early_stopping_patience: 2
eval_max_new_tokens: 128
eval_steps: 100
eval_table_size: null
flash_attention: false
fp16: null
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 8
gradient_checkpointing: true
group_by_length: false
hub_model_id: Alphatao/a2fadffe-74bd-40df-9b59-79a4857925a7
hub_repo: null
hub_strategy: checkpoint
hub_token: null
learning_rate: 0.0002
load_best_model_at_end: true
load_in_4bit: false
load_in_8bit: false
local_rank: null
logging_steps: 1
lora_alpha: 32
lora_dropout: 0.05
lora_fan_in_fan_out: null
lora_model_dir: null
lora_r: 16
lora_target_linear: true
lora_target_modules:
- q_proj
- k_proj
lr_scheduler: cosine
max_grad_norm: 1.0
max_steps: 4140
micro_batch_size: 4
mlflow_experiment_name: /tmp/f1bfae7be46056e8_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 2
optimizer: adamw_bnb_8bit
output_dir: miner_id_24
pad_to_sequence_len: true
resume_from_checkpoint: null
s2_attention: null
sample_packing: false
save_steps: 100
sequence_len: 1024
special_tokens:
  pad_token: <|endoftext|>
strict: false
tf32: true
tokenizer_type: AutoTokenizer
train_on_inputs: false
trust_remote_code: true
val_set_size: 0.05
wandb_entity: null
wandb_mode: online
wandb_name: ff228523-b38e-4081-8a47-fba2a3a7734a
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: ff228523-b38e-4081-8a47-fba2a3a7734a
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

a2fadffe-74bd-40df-9b59-79a4857925a7

This model is a fine-tuned version of EleutherAI/gpt-neo-125m on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.4955

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 0.0002
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 32
• optimizer: Use OptimizerNames.ADAMW_BNB with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 10
• training_steps: 4140

Training results

Training Loss	Epoch	Step	Validation Loss
14.944	0.0004	1	1.8463
13.3083	0.0410	100	1.6581
12.6572	0.0819	200	1.6218
12.1718	0.1229	300	1.6031
13.1425	0.1638	400	1.5900
11.7675	0.2048	500	1.5797
12.2515	0.2458	600	1.5709
12.7135	0.2867	700	1.5630
12.4054	0.3277	800	1.5568
11.8353	0.3686	900	1.5502
12.0527	0.4096	1000	1.5448
12.2754	0.4506	1100	1.5402
12.0864	0.4915	1200	1.5360
12.0309	0.5325	1300	1.5317
10.92	0.5734	1400	1.5285
12.7007	0.6144	1500	1.5247
11.1532	0.6554	1600	1.5218
12.6976	0.6963	1700	1.5187
11.9156	0.7373	1800	1.5166
12.6085	0.7782	1900	1.5139
12.4087	0.8192	2000	1.5118
12.1581	0.8602	2100	1.5099
11.9825	0.9011	2200	1.5079
11.2843	0.9421	2300	1.5060
12.8118	0.9831	2400	1.5049
11.1252	1.0240	2500	1.5034
10.9378	1.0650	2600	1.5022
12.2633	1.1059	2700	1.5009
11.5464	1.1469	2800	1.5001
11.6295	1.1879	2900	1.4994
12.1325	1.2288	3000	1.4985
11.571	1.2698	3100	1.4978
12.381	1.3107	3200	1.4973
11.4236	1.3517	3300	1.4968
12.2288	1.3927	3400	1.4964
12.2337	1.4336	3500	1.4961
12.5231	1.4746	3600	1.4958
11.0633	1.5155	3700	1.4957
12.0329	1.5565	3800	1.4956
11.9344	1.5975	3900	1.4956
12.7316	1.6384	4000	1.4955
12.2778	1.6794	4100	1.4955

Framework versions

• PEFT 0.13.2
• Transformers 4.46.0
• Pytorch 2.5.0+cu124
• Datasets 3.0.1
• Tokenizers 0.20.1