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:
  - 2ed3e6fceaadcf92_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/2ed3e6fceaadcf92_train_data.json
  type:
    field_instruction: instruction
    field_output: response
    format: '{instruction}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
device_map:
  ? ''
  : 0,1,2,3,4,5,6,7
early_stopping_patience: 2
eval_max_new_tokens: 128
eval_steps: 100
eval_table_size: null
flash_attention: false
gradient_accumulation_steps: 8
gradient_checkpointing: true
group_by_length: false
hub_model_id: Alphatao/e9d6d562-a8a2-4cba-a9a9-a63aac7ea772
hub_repo: null
hub_strategy: null
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: null
lr_scheduler: cosine
max_grad_norm: 1.0
max_steps: 3600
micro_batch_size: 4
mlflow_experiment_name: /tmp/2ed3e6fceaadcf92_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: 0.025100401606425703
wandb_entity: null
wandb_mode: online
wandb_name: ec6075d1-7d73-42f5-a76a-d714ba1db9c4
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: ec6075d1-7d73-42f5-a76a-d714ba1db9c4
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

e9d6d562-a8a2-4cba-a9a9-a63aac7ea772

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: 1.3511

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: 3600

Training results

Training Loss	Epoch	Step	Validation Loss
13.1793	0.0002	1	1.5766
11.231	0.0165	100	1.4597
10.4955	0.0330	200	1.4353
11.3411	0.0494	300	1.4223
10.7846	0.0659	400	1.4131
11.6785	0.0824	500	1.4054
10.8221	0.0989	600	1.3992
10.7452	0.1153	700	1.3938
10.3698	0.1318	800	1.3888
12.2309	0.1483	900	1.3846
12.0698	0.1648	1000	1.3808
10.4695	0.1813	1100	1.3777
11.0436	0.1977	1200	1.3749
9.6067	0.2142	1300	1.3727
12.0766	0.2307	1400	1.3698
12.1263	0.2472	1500	1.3673
10.48	0.2636	1600	1.3652
11.351	0.2801	1700	1.3636
9.9537	0.2966	1800	1.3619
11.0333	0.3131	1900	1.3600
10.0042	0.3296	2000	1.3589
10.4926	0.3460	2100	1.3577
10.4749	0.3625	2200	1.3565
11.3601	0.3790	2300	1.3557
10.7283	0.3955	2400	1.3549
11.6171	0.4119	2500	1.3541
12.0398	0.4284	2600	1.3532
10.4231	0.4449	2700	1.3529
11.7707	0.4614	2800	1.3523
11.7205	0.4779	2900	1.3519
11.2512	0.4943	3000	1.3515
10.9363	0.5108	3100	1.3514
10.654	0.5273	3200	1.3513
10.9013	0.5438	3300	1.3512
11.0595	0.5602	3400	1.3511
11.3651	0.5767	3500	1.3511
10.8299	0.5932	3600	1.3511

Framework versions

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