See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: NousResearch/CodeLlama-7b-hf
bf16: auto
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - 45fb2d361254b178_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/45fb2d361254b178_train_data.json
  type:
    field_input: counter_longer
    field_instruction: counter_statement
    field_output: question
    format: '{instruction} {input}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
deepspeed: null
early_stopping_patience: null
eval_max_new_tokens: 128
eval_table_size: null
evals_per_epoch: 4
flash_attention: false
fp16: null
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 16
gradient_checkpointing: false
group_by_length: false
hub_model_id: error577/58b9523a-8576-4309-80c7-060f2d6bf699
hub_repo: null
hub_strategy: checkpoint
hub_token: null
learning_rate: 0.0002
load_in_4bit: true
load_in_8bit: false
local_rank: null
logging_steps: 1
lora_alpha: 16
lora_dropout: 0.05
lora_fan_in_fan_out: null
lora_model_dir: null
lora_r: 4
lora_target_linear: true
lr_scheduler: cosine
max_steps: 20
micro_batch_size: 1
mlflow_experiment_name: /tmp/45fb2d361254b178_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 4
optimizer: adamw_bnb_8bit
output_dir: miner_id_24
pad_to_sequence_len: true
resume_from_checkpoint: null
s2_attention: null
sample_packing: false
saves_per_epoch: 4
sequence_len: 256
special_tokens:
  pad_token: </s>
strict: false
tf32: false
tokenizer_type: AutoTokenizer
train_on_inputs: false
trust_remote_code: true
val_set_size: 0.05
wandb_entity: null
wandb_mode: online
wandb_name: 6344f2fc-fa2b-4848-99cc-8281347a9bf0
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: 6344f2fc-fa2b-4848-99cc-8281347a9bf0
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

58b9523a-8576-4309-80c7-060f2d6bf699

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

• Loss: 0.3307

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: 1
• eval_batch_size: 1
• seed: 42
• gradient_accumulation_steps: 16
• total_train_batch_size: 16
• 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: 20

Training results

Training Loss	Epoch	Step	Validation Loss
43.4533	0.0015	1	2.6565
39.2663	0.0030	2	2.6556
39.5743	0.0059	4	2.6245
37.8648	0.0089	6	2.4578
36.6476	0.0119	8	1.9450
23.4278	0.0148	10	1.1998
11.9862	0.0178	12	0.5601
4.9234	0.0208	14	0.3981
9.4431	0.0237	16	0.3685
1.5801	0.0267	18	0.3404
5.968	0.0297	20	0.3307

Framework versions

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