See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: NousResearch/CodeLlama-7b-hf-flash
bf16: true
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - bb63f2845e2675bc_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/bb63f2845e2675bc_train_data.json
  type:
    field_input: sub_topic
    field_instruction: question
    field_output: answer
    format: '{instruction} {input}'
    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: 400
eval_table_size: null
flash_attention: false
gradient_accumulation_steps: 4
gradient_checkpointing: true
group_by_length: false
hub_model_id: Alphatao/83c0f599-dd89-4ae1-84b6-b5fe4a0bace8
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: 3684
micro_batch_size: 2
mlflow_experiment_name: /tmp/bb63f2845e2675bc_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: 400
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: 0.04
wandb_entity: null
wandb_mode: online
wandb_name: 709a6d28-9f8a-4848-8614-7ea87b70604a
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: 709a6d28-9f8a-4848-8614-7ea87b70604a
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

83c0f599-dd89-4ae1-84b6-b5fe4a0bace8

This model is a fine-tuned version of NousResearch/CodeLlama-7b-hf-flash on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.3157

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: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 8
• 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: 3684

Training results

Training Loss	Epoch	Step	Validation Loss
2.6698	0.0002	1	0.6571
1.3487	0.0678	400	0.4049
1.0191	0.1355	800	0.3768
1.5641	0.2033	1200	0.3599
1.3864	0.2710	1600	0.3462
1.3802	0.3388	2000	0.3342
1.5108	0.4065	2400	0.3258
0.9273	0.4743	2800	0.3197
1.1166	0.5420	3200	0.3166
0.7929	0.6098	3600	0.3157

Framework versions

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