See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: unsloth/Llama-3.2-3B-Instruct
bf16: auto
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - MATH-Hard_train_data.json
  ds_type: json
  path: /workspace/input_data/MATH-Hard_train_data.json
  type:
    field_input: solution
    field_instruction: type
    field_output: problem
    system_format: '{system}'
    system_prompt: ''
debug: null
deepspeed: null
early_stopping_patience: 1
eval_max_new_tokens: 128
eval_steps: 5
eval_table_size: null
flash_attention: true
fp16: null
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 4
gradient_checkpointing: true
group_by_length: false
hours_to_complete: 1
hub_model_id: besimray/miner1_bdb2cb48-85f5-4cf3-b680-0d740624ff07_1731086176
hub_strategy: checkpoint
hub_token: null
learning_rate: 0.0002
load_in_4bit: false
load_in_8bit: true
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: 32
lora_target_linear: true
lr_scheduler: cosine
max_steps: 500
micro_batch_size: 2
mlflow_experiment_name: /tmp/MATH-Hard_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
save_steps: 10
save_strategy: steps
sequence_len: 4096
started_at: '2024-11-08T17:16:16.598479'
strict: false
tf32: false
tokenizer_type: AutoTokenizer
train_on_inputs: false
trust_remote_code: true
val_set_size: 0.05
wandb_entity: besimray24-rayon
wandb_mode: online
wandb_project: Public_TuningSN
wandb_run: miner_id_24
wandb_runid: bdb2cb48-85f5-4cf3-b680-0d740624ff07
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

miner1_bdb2cb48-85f5-4cf3-b680-0d740624ff07_1731086176

This model is a fine-tuned version of unsloth/Llama-3.2-3B-Instruct on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.8078

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: 500

Training results

Training Loss	Epoch	Step	Validation Loss
1.612	0.0024	1	1.3749
1.1252	0.0119	5	1.3356
0.3899	0.0239	10	1.0437
1.4463	0.0358	15	0.9187
0.7956	0.0477	20	0.8748
0.9788	0.0597	25	0.8622
1.1846	0.0716	30	0.8492
0.8456	0.0835	35	0.8385
0.6094	0.0955	40	0.8294
0.5703	0.1074	45	0.8229
0.7308	0.1193	50	0.8160
0.86	0.1313	55	0.8154
0.7686	0.1432	60	0.8121
0.556	0.1551	65	0.8097
1.2009	0.1671	70	0.8047
0.7956	0.1790	75	0.8078

Framework versions

• PEFT 0.13.2
• Transformers 4.46.1
• Pytorch 2.3.1+cu121
• Datasets 3.0.1
• Tokenizers 0.20.3