See axolotl config

axolotl version: 0.6.0

adapter: lora
base_model: bigcode/starcoder2-3b
bf16: true
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - 012ab4813cc99fb8_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/012ab4813cc99fb8_train_data.json
  type:
    field_input: evidence
    field_instruction: question
    field_output: SQL
    format: '{instruction} {input}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
deepspeed: null
device_map: auto
do_eval: true
early_stopping_patience: 5
eval_batch_size: 4
eval_max_new_tokens: 128
eval_steps: 50
eval_table_size: null
evals_per_epoch: null
flash_attention: false
fp16: false
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 2
gradient_checkpointing: true
group_by_length: true
hub_model_id: dabrown/3ca3719e-ac62-4291-8edc-c53670bea96a
hub_repo: null
hub_strategy: checkpoint
hub_token: null
learning_rate: 0.0001
load_in_4bit: false
load_in_8bit: false
local_rank: null
logging_steps: 1
lora_alpha: 128
lora_dropout: 0.2
lora_fan_in_fan_out: false
lora_inference_mode: true
lora_model_dir: null
lora_modules_to_save:
- lm_head
lora_r: 64
lora_target_linear: true
lr_scheduler: cosine
max_grad_norm: 1.0
max_memory:
  0: 75GB
micro_batch_size: 4
mlflow_experiment_name: /tmp/012ab4813cc99fb8_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 10
optim_args:
  adam_beta1: 0.9
  adam_beta2: 0.95
  adam_epsilon: 1e-5
optimizer: adamw_bnb_8bit
output_dir: miner_id_24
pad_to_sequence_len: true
peft_use_rslora: true
resume_from_checkpoint: null
s2_attention: null
sample_packing: false
save_steps: 50
saves_per_epoch: null
sequence_len: 1024
shuffle_merged_datasets: true
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: b1e23278-252e-44d7-9491-1b28d344421c
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: b1e23278-252e-44d7-9491-1b28d344421c
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

3ca3719e-ac62-4291-8edc-c53670bea96a

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: 0.3206

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: 0.0001
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 8
• optimizer: Use adamw_bnb_8bit with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=adam_beta1=0.9,adam_beta2=0.95,adam_epsilon=1e-5
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 10
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss
0.5298	0.2525	50	0.4032
0.5869	0.5051	100	0.3490
0.445	0.7576	150	0.3094
0.3591	1.0101	200	0.3043
0.3098	1.2626	250	0.3415
0.3208	1.5152	300	0.3217
0.2425	1.7677	350	0.3043
0.1494	2.0202	400	0.3094
0.1766	2.2727	450	0.3060
0.215	2.5253	500	0.3022
0.1314	2.7778	550	0.3130
0.1838	3.0303	600	0.3147
0.1067	3.2828	650	0.3202
0.1491	3.5354	700	0.3042
0.1563	3.7879	750	0.3206

Framework versions

• PEFT 0.14.0
• Transformers 4.46.3
• Pytorch 2.5.1+cu124
• Datasets 3.1.0
• Tokenizers 0.20.3