See axolotl config

axolotl version: 0.10.0.dev0

adapter: lora
base_model: TinyLlama/TinyLlama-1.1B-Chat-v0.6
bf16: true
chat_template: llama3
datasets:
- data_files:
  - 57d47e30a3fd3c23_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: 2
gradient_checkpointing: true
group_by_length: true
hub_model_id: apriasmoro/b8556b3c-8e5b-4968-bdbb-f67c199efafa
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: 71
micro_batch_size: 8
mlflow_experiment_name: /tmp/57d47e30a3fd3c23_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: 10
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: cbc2c6ab-186e-46fb-ad26-97569d03f5e2
wandb_project: Gradients-On-Demand
wandb_run: apriasmoro
wandb_runid: cbc2c6ab-186e-46fb-ad26-97569d03f5e2
warmup_steps: 100
weight_decay: 0.01

b8556b3c-8e5b-4968-bdbb-f67c199efafa

This model is a fine-tuned version of TinyLlama/TinyLlama-1.1B-Chat-v0.6 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.2711

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: 8
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 8
• gradient_accumulation_steps: 2
• total_train_batch_size: 128
• total_eval_batch_size: 64
• 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: 71

Training results

Training Loss	Epoch	Step	Validation Loss
No log	0.0083	1	2.0375
1.9147	0.1	12	2.0346
1.9885	0.2	24	2.0063
1.9495	0.3	36	1.8368
1.7864	0.4	48	1.5685
1.3824	0.5	60	1.2711

Framework versions

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