See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: microsoft/Phi-3-mini-128k-instruct
bf16: true
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - f519e55b9fa5e216_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/f519e55b9fa5e216_train_data.json
  type:
    field_input: intent
    field_instruction: instruction
    field_output: response
    format: '{instruction} {input}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
device_map:
  ? ''
  : 0,1,2,3,4,5,6,7
early_stopping_patience: 2
eval_max_new_tokens: 128
eval_steps: 100
eval_table_size: null
flash_attention: false
gradient_accumulation_steps: 8
gradient_checkpointing: true
group_by_length: false
hub_model_id: Alphatao/73bb6bbb-83b4-4a28-98fd-5753f0f5a11d
hub_repo: null
hub_strategy: null
hub_token: null
learning_rate: 0.0002
load_best_model_at_end: true
load_in_4bit: false
load_in_8bit: false
local_rank: null
logging_steps: 1
lora_alpha: 128
lora_dropout: 0.1
lora_fan_in_fan_out: null
lora_model_dir: null
lora_r: 64
lora_target_linear: true
lora_target_modules:
- q_proj
- k_proj
- v_proj
- o_proj
lr_scheduler: cosine
max_grad_norm: 1.0
max_steps: 2040
micro_batch_size: 4
mlflow_experiment_name: /tmp/f519e55b9fa5e216_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 2
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: 100
sequence_len: 1024
strict: false
tf32: true
tokenizer_type: AutoTokenizer
train_on_inputs: false
trust_remote_code: true
val_set_size: 0.020094686161191533
wandb_entity: null
wandb_mode: online
wandb_name: d6152391-97e5-4079-b599-8fa049291a62
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: d6152391-97e5-4079-b599-8fa049291a62
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

73bb6bbb-83b4-4a28-98fd-5753f0f5a11d

This model is a fine-tuned version of microsoft/Phi-3-mini-128k-instruct on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.2567

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
• gradient_accumulation_steps: 8
• total_train_batch_size: 32
• 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: 2040

Training results

Training Loss	Epoch	Step	Validation Loss
3.462	0.0001	1	0.5078
2.1475	0.0131	100	0.2966
2.1053	0.0262	200	0.2892
2.2196	0.0394	300	0.2850
2.2406	0.0525	400	0.2813
2.0445	0.0656	500	0.2786
2.2754	0.0787	600	0.2756
2.5719	0.0919	700	0.2734
2.4255	0.1050	800	0.2715
2.6879	0.1181	900	0.2692
1.9891	0.1312	1000	0.2669
2.3204	0.1444	1100	0.2650
2.1741	0.1575	1200	0.2632
2.5809	0.1706	1300	0.2618
2.0244	0.1837	1400	0.2603
2.5022	0.1969	1500	0.2592
2.3195	0.2100	1600	0.2581
2.1718	0.2231	1700	0.2575
2.2811	0.2362	1800	0.2570
1.6784	0.2494	1900	0.2568
2.1885	0.2625	2000	0.2567

Framework versions

• PEFT 0.13.2
• Transformers 4.46.0
• Pytorch 2.5.0+cu124
• Datasets 3.0.1
• Tokenizers 0.20.1