See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: heegyu/WizardVicuna-open-llama-3b-v2
bf16: true
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - c6c4a1836b6ca46f_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/c6c4a1836b6ca46f_train_data.json
  type:
    field_instruction: instruction
    field_output: output
    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: true
fp16: null
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 8
gradient_checkpointing: true
group_by_length: false
hub_model_id: Alphatao/194985da-ea25-4ebf-b6f2-4eb986f88fb8
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
- v_proj
lr_scheduler: cosine
max_grad_norm: 1.0
max_steps: 2040
micro_batch_size: 4
mlflow_experiment_name: /tmp/c6c4a1836b6ca46f_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: 0.045068594400677835
wandb_entity: null
wandb_mode: online
wandb_name: 6d7da5ab-5de4-4404-a75f-58ba9fdee6a6
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: 6d7da5ab-5de4-4404-a75f-58ba9fdee6a6
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

194985da-ea25-4ebf-b6f2-4eb986f88fb8

This model is a fine-tuned version of heegyu/WizardVicuna-open-llama-3b-v2 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.2330

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: 2040

Training results

Training Loss	Epoch	Step	Validation Loss
1.8914	0.0003	1	1.6873
1.234	0.0302	100	1.3915
1.2987	0.0604	200	1.3537
1.3263	0.0906	300	1.3335
1.3214	0.1208	400	1.3190
1.3969	0.1510	500	1.3080
1.2744	0.1812	600	1.2942
1.2006	0.2114	700	1.2864
1.2436	0.2416	800	1.2758
1.2415	0.2718	900	1.2692
1.1909	0.3020	1000	1.2629
1.2246	0.3323	1100	1.2566
1.3166	0.3625	1200	1.2513
1.1357	0.3927	1300	1.2469
1.3915	0.4229	1400	1.2432
1.1805	0.4531	1500	1.2396
1.2176	0.4833	1600	1.2368
1.2613	0.5135	1700	1.2348
1.2448	0.5437	1800	1.2337
1.2321	0.5739	1900	1.2332
1.2082	0.6041	2000	1.2330

Framework versions

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