See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: EleutherAI/gpt-neo-1.3B
bf16: true
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - e5625f7766855655_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/e5625f7766855655_train_data.json
  type:
    field_instruction: chat
    field_output: text
    format: '{instruction}'
    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: romainnn/f083a5bd-ba97-48d4-980a-c85badd3b363
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: 2346
micro_batch_size: 4
mlflow_experiment_name: /tmp/e5625f7766855655_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: 0.04438132433871827
wandb_entity: null
wandb_mode: online
wandb_name: 6bb1ae2c-eec7-426b-8297-d030ba828c03
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: 6bb1ae2c-eec7-426b-8297-d030ba828c03
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

f083a5bd-ba97-48d4-980a-c85badd3b363

This model is a fine-tuned version of EleutherAI/gpt-neo-1.3B on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.0071

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: 2346

Training results

Training Loss	Epoch	Step	Validation Loss
5.119	0.0003	1	0.6604
0.1114	0.0297	100	0.0141
0.078	0.0594	200	0.0110
0.1317	0.0892	300	0.0098
0.0564	0.1189	400	0.0093
0.0795	0.1486	500	0.0090
0.0728	0.1783	600	0.0088
0.0607	0.2081	700	0.0087
0.0625	0.2378	800	0.0086
0.0647	0.2675	900	0.0083
0.0724	0.2972	1000	0.0081
0.0765	0.3270	1100	0.0079
0.0637	0.3567	1200	0.0078
0.0796	0.3864	1300	0.0077
0.042	0.4161	1400	0.0076
0.0556	0.4458	1500	0.0075
0.053	0.4756	1600	0.0074
0.0722	0.5053	1700	0.0074
0.0631	0.5350	1800	0.0073
0.0485	0.5647	1900	0.0072
0.0627	0.5945	2000	0.0072
0.0765	0.6242	2100	0.0072
0.1001	0.6539	2200	0.0072
0.0643	0.6836	2300	0.0071

Framework versions

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