Lllama_AHS_V_7.0 / README.md
Tatvajsh's picture
Update README.md
a2898dc
metadata
license: apache-2.0
base_model: openlm-research/open_llama_3b_v2
tags:
  - generated_from_trainer
model-index:
  - name: working
    results: []

working

This model is a fine-tuned version of openlm-research/open_llama_3b_v2 on the None dataset.

Model description

training_arguments = TrainingArguments( per_device_train_batch_size=8, num_train_epochs=10, learning_rate=3e-5, gradient_accumulation_steps=2, optim="adamw_hf", fp16=True, logging_steps=1, # debug=True, output_dir="/kaggle/Tatvajsh/Lllama_AHS_V_7.0/" # warmup_steps=100, )

trainer = SFTTrainer( model=model, tokenizer=tokenizer, train_dataset=dataset, dataset_text_field="text", peft_config=lora_config, max_seq_length=512, args=training_arguments,

packing=True,#change

)

trainer.train()

EPOCHS=[30-50]

from peft import LoraConfig, get_peft_model

lora_config = LoraConfig( r=16, lora_alpha=64, target_modules=['base_layer','gate_proj', 'v_proj','up_proj','down_proj','q_proj','k_proj','o_proj'], lora_dropout=0.05, bias="none", task_type="CAUSAL_LM" )

def generate_prompt(row) -> str: prompt=f""" Below is an instruction that describes a task. Write a response that appropriately completes the request.

### Instruction:

{row['Instruction']} 

### Response:

{row['Answer']}  

### End
"""
return prompt

WHEN THE TRAINING LOSS IN NOT REDUCING THEN TRY SETTING FOR LESSER VALUE OF LEARNING RATE I.E. 2E-5 TO 3E-5,ETC. 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: 3e-05
  • train_batch_size: 8
  • eval_batch_size: 8
  • seed: 42
  • gradient_accumulation_steps: 2
  • total_train_batch_size: 16
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 10
  • mixed_precision_training: Native AMP

Training results

Framework versions

  • Transformers 4.35.0.dev0
  • Pytorch 2.0.0
  • Datasets 2.1.0
  • Tokenizers 0.14.1