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:
  - Omni-MATH_train_data.json
  ds_type: json
  path: /workspace/input_data/Omni-MATH_train_data.json
  type:
    field_input: domain
    field_instruction: problem
    field_output: solution
    system_format: '{system}'
    system_prompt: ''
debug: null
deepspeed: null
early_stopping_patience: 3
eval_max_new_tokens: 128
eval_steps: 10
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
hub_model_id: besimray/miner_id_3_7a3ce630-aa05-41fd-9ccb-55fbe77a4f6b_1729737185
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: 1000
micro_batch_size: 5
mlflow_experiment_name: /tmp/Omni-MATH_train_data.json
model_type: LlamaForCausalLM
num_epochs: 5
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
strict: false
tf32: false
tokenizer_type: AutoTokenizer
train_on_inputs: false
val_set_size: 0.05
wandb_entity: besimray24-rayon
wandb_mode: online
wandb_project: Public_TuningSN
wandb_run: miner_id_24
wandb_runid: 7a3ce630-aa05-41fd-9ccb-55fbe77a4f6b
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

miner_id_3_7a3ce630-aa05-41fd-9ccb-55fbe77a4f6b_1729737185

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: 0.9668

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: 5
• eval_batch_size: 5
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 20
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 10
• training_steps: 1000

Training results

Training Loss	Epoch	Step	Validation Loss
1.0856	0.0049	1	1.1341
0.9932	0.0486	10	1.0513
0.9157	0.0972	20	1.0124
0.9999	0.1458	30	0.9980
0.8839	0.1944	40	0.9909
1.0734	0.2430	50	0.9855
0.9152	0.2916	60	0.9828
1.1024	0.3402	70	0.9797
0.938	0.3888	80	0.9775
0.8866	0.4374	90	0.9766
0.9912	0.4860	100	0.9744
0.9395	0.5346	110	0.9725
0.9197	0.5832	120	0.9711
0.8917	0.6318	130	0.9689
0.9746	0.6804	140	0.9663
0.8681	0.7290	150	0.9668
0.8568	0.7776	160	0.9672
1.062	0.8262	170	0.9668

Framework versions

• PEFT 0.13.2
• Transformers 4.45.2
• Pytorch 2.4.1+cu124
• Datasets 3.0.1
• Tokenizers 0.20.1