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:
  - 70a3ad0aa1ce028e_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/70a3ad0aa1ce028e_train_data.json
  type:
    field_instruction: instruction
    field_output: chosen_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: true
gradient_accumulation_steps: 8
gradient_checkpointing: true
group_by_length: false
hub_model_id: Alphatao/2aa189bb-435b-4119-b1f2-d6be4664ac5f
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: 128
lora_dropout: 0.3
lora_fan_in_fan_out: null
lora_model_dir: null
lora_r: 64
lora_target_linear: true
lora_target_modules:
- q_proj
- k_proj
- v_proj
- o_proj
- gate_proj
- down_proj
- up_proj
lr_scheduler: cosine
max_grad_norm: 1.0
max_steps: 2040
micro_batch_size: 4
mlflow_experiment_name: /tmp/70a3ad0aa1ce028e_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: 0.05
wandb_entity: null
wandb_mode: online
wandb_name: 694d3442-1f93-4c0f-9b2a-dc0a141cb41e
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: 694d3442-1f93-4c0f-9b2a-dc0a141cb41e
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

2aa189bb-435b-4119-b1f2-d6be4664ac5f

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: 0.9911

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
1.4522	0.0005	1	1.3640
1.1473	0.0485	100	1.1701
1.1545	0.0971	200	1.1438
1.1014	0.1456	300	1.1045
1.2081	0.1941	400	1.0869
1.1507	0.2427	500	1.0752
1.0523	0.2912	600	1.0613
1.1854	0.3397	700	1.0523
1.0858	0.3883	800	1.0430
1.1101	0.4368	900	1.0353
0.8973	0.4853	1000	1.0274
0.9486	0.5339	1100	1.0197
1.029	0.5824	1200	1.0131
0.9404	0.6310	1300	1.0072
0.9105	0.6795	1400	1.0029
1.1133	0.7280	1500	0.9987
1.0234	0.7766	1600	0.9957
1.0717	0.8251	1700	0.9935
0.9703	0.8736	1800	0.9920
1.2598	0.9222	1900	0.9913
0.9952	0.9707	2000	0.9911

Framework versions

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