See axolotl config

axolotl version: 0.4.1

adapter: lora
auto_find_batch_size: true
base_model: bigcode/starcoder2-3b
bf16: auto
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - 740553afd00c7510_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/740553afd00c7510_train_data.json
  type:
    field_input: context
    field_instruction: question-X
    field_output: answer-Y
    format: '{instruction} {input}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
deepspeed: null
do_eval: true
early_stopping_patience: 3
eval_max_new_tokens: 128
eval_steps: 50
evals_per_epoch: null
flash_attention: true
fp16: false
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 2
gradient_checkpointing: false
group_by_length: true
hub_model_id: lesso10/392a6c68-053d-4422-9b25-ad3be3dfea2d
hub_repo: null
hub_strategy: checkpoint
hub_token: null
learning_rate: 0.00021
load_in_4bit: false
load_in_8bit: false
local_rank: null
logging_steps: 10
lora_alpha: 32
lora_dropout: 0.05
lora_fan_in_fan_out: null
lora_model_dir: null
lora_r: 16
lora_target_linear: true
lr_scheduler: cosine
max_grad_norm: 1.0
max_steps: 500
micro_batch_size: 4
mlflow_experiment_name: /tmp/740553afd00c7510_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 1
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: 50
saves_per_epoch: null
seed: 100
sequence_len: 512
special_tokens:
  pad_token: <|endoftext|>
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: daa7dc44-d862-4c53-9590-6306c3403b65
wandb_project: 10a
wandb_run: your_name
wandb_runid: daa7dc44-d862-4c53-9590-6306c3403b65
warmup_steps: 50
weight_decay: 0.0
xformers_attention: null

392a6c68-053d-4422-9b25-ad3be3dfea2d

This model is a fine-tuned version of bigcode/starcoder2-3b on the None dataset. It achieves the following results on the evaluation set:

• Loss: 2.0223

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: 0.00021
• train_batch_size: 4
• eval_batch_size: 4
• seed: 100
• gradient_accumulation_steps: 2
• 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: 50
• training_steps: 500

Training results

Training Loss	Epoch	Step	Validation Loss
No log	0.0002	1	4.5463
6.7133	0.0123	50	2.2650
6.0063	0.0247	100	2.2393
5.3395	0.0370	150	2.1622
4.9373	0.0494	200	2.1393
4.7686	0.0617	250	2.1041
4.8419	0.0740	300	2.0924
5.0222	0.0864	350	2.0310
4.4282	0.0987	400	2.0324
4.2734	0.1111	450	2.0221
4.5621	0.1234	500	2.0223

Framework versions

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