See axolotl config

axolotl version: 0.10.0.dev0

adapter: lora
base_model: unsloth/Qwen2.5-3B
bf16: true
chat_template: llama3
datasets:
- data_files:
  - 9b229213575401f4_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/
  type:
    field_instruction: instruct
    field_output: output
    format: '{instruction}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
eval_max_new_tokens: 256
evals_per_epoch: 2
flash_attention: false
fp16: false
gradient_accumulation_steps: 1
gradient_checkpointing: true
group_by_length: true
hub_model_id: segopecelus/d1b844dc-f2d6-4d65-9484-a7e5dd74533d
learning_rate: 0.0002
logging_steps: 10
lora_alpha: 16
lora_dropout: 0.05
lora_fan_in_fan_out: false
lora_r: 8
lora_target_linear: true
lr_scheduler: cosine
max_steps: 45
micro_batch_size: 4
mlflow_experiment_name: /tmp/9b229213575401f4_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 3
optimizer: adamw_bnb_8bit
output_dir: miner_id_24
pad_to_sequence_len: true
sample_packing: false
save_steps: 34
sequence_len: 2048
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: 999e249f-6b05-4a37-9bc6-b4556645f48a
wandb_project: Gradients-On-Demand
wandb_run: apriasmoro
wandb_runid: 999e249f-6b05-4a37-9bc6-b4556645f48a
warmup_steps: 100
weight_decay: 0.01

d1b844dc-f2d6-4d65-9484-a7e5dd74533d

This model is a fine-tuned version of unsloth/Qwen2.5-3B on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 3.0708

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: 4
• eval_batch_size: 4
• seed: 42
• 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: 100
• training_steps: 45

Training results

Training Loss	Epoch	Step	Validation Loss
No log	0.0005	1	3.4306
No log	0.0043	8	3.3986
1.561	0.0085	16	3.4275
2.0519	0.0128	24	3.4316
2.9312	0.0170	32	3.3414
3.0088	0.0213	40	3.0708

Framework versions

• PEFT 0.15.2
• Transformers 4.51.3
• Pytorch 2.5.1+cu124
• Datasets 3.5.1
• Tokenizers 0.21.1