See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: HuggingFaceH4/tiny-random-LlamaForCausalLM
bf16: true
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - 0e2a206b6cbe63d9_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/0e2a206b6cbe63d9_train_data.json
  type:
    field_instruction: prompt
    field_output: chosen
    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: true
gradient_accumulation_steps: 8
gradient_checkpointing: true
group_by_length: false
hub_model_id: Alphatao/90dbfeae-95d3-47a2-a988-98c5906bae01
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:
- q_proj
- k_proj
- v_proj
- o_proj
- down_proj
- up_proj
lr_scheduler: cosine
max_grad_norm: 1.0
max_steps: 3600
micro_batch_size: 4
mlflow_experiment_name: /tmp/0e2a206b6cbe63d9_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: 2048
special_tokens:
  pad_token: </s>
strict: false
tf32: true
tokenizer_type: AutoTokenizer
train_on_inputs: false
trust_remote_code: true
val_set_size: 0.03361547925588775
wandb_entity: null
wandb_mode: online
wandb_name: 65ecc54e-1ce1-46d0-8d8f-a58fd50f5f0f
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: 65ecc54e-1ce1-46d0-8d8f-a58fd50f5f0f
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

90dbfeae-95d3-47a2-a988-98c5906bae01

This model is a fine-tuned version of HuggingFaceH4/tiny-random-LlamaForCausalLM on the None dataset. It achieves the following results on the evaluation set:

• Loss: 10.3113

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
10.3673	0.0002	1	10.3688
10.3363	0.0223	100	10.3367
10.3308	0.0445	200	10.3281
10.324	0.0668	300	10.3221
10.3186	0.0890	400	10.3186
10.3196	0.1113	500	10.3180
10.3207	0.1336	600	10.3175
10.321	0.1558	700	10.3171
10.3129	0.1781	800	10.3167
10.3111	0.2004	900	10.3158
10.3192	0.2226	1000	10.3152
10.3219	0.2449	1100	10.3147
10.3197	0.2671	1200	10.3142
10.3143	0.2894	1300	10.3139
10.3172	0.3117	1400	10.3135
10.315	0.3339	1500	10.3132
10.3181	0.3562	1600	10.3128
10.3104	0.3785	1700	10.3125
10.3086	0.4007	1800	10.3123
10.3091	0.4230	1900	10.3121
10.3141	0.4452	2000	10.3119
10.314	0.4675	2100	10.3118
10.3158	0.4898	2200	10.3117
10.3147	0.5120	2300	10.3117
10.3192	0.5343	2400	10.3116
10.3062	0.5565	2500	10.3115
10.3046	0.5788	2600	10.3115
10.3184	0.6011	2700	10.3114
10.3104	0.6233	2800	10.3114
10.3105	0.6456	2900	10.3114
10.3174	0.6679	3000	10.3114
10.3149	0.6901	3100	10.3113
10.3201	0.7124	3200	10.3113
10.3111	0.7346	3300	10.3113
10.3115	0.7569	3400	10.3113
10.3168	0.7792	3500	10.3113
10.3123	0.8014	3600	10.3113

Framework versions

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