Training procedure

The following bitsandbytes quantization config was used during training:

• load_in_8bit: False
• load_in_4bit: True
• llm_int8_threshold: 6.0
• llm_int8_skip_modules: None
• llm_int8_enable_fp32_cpu_offload: False
• llm_int8_has_fp16_weight: False
• bnb_4bit_quant_type: nf4
• bnb_4bit_use_double_quant: True
• bnb_4bit_compute_dtype: bfloat16

Framework versions

• PEFT 0.4.0

How to use:

!pip install transformers peft accelerate bitsandbytes trl safetensors

from huggingface_hub import notebook_login
notebook_login()

import torch
from peft import AutoPeftModelForCausalLM, get_peft_config, PeftModel, PeftConfig, get_peft_model, LoraConfig, TaskType
from transformers import AutoTokenizer

peft_model_id = "akdeniz27/llama-2-7b-hf-qlora-dolly15k-turkish"
config = PeftConfig.from_pretrained(peft_model_id)
# load base LLM model and tokenizer
model = AutoPeftModelForCausalLM.from_pretrained(
    peft_model_id,
    low_cpu_mem_usage=True,
    torch_dtype=torch.float16,
    load_in_4bit=True,
)
tokenizer = AutoTokenizer.from_pretrained(config.base_model_name_or_path)

prompt = "..."

input_ids = tokenizer(prompt, return_tensors="pt", truncation=True).input_ids.cuda()

outputs = model.generate(input_ids=input_ids, max_new_tokens=100, do_sample=True, top_p=0.9,temperature=0.9)