suriya7/Gemma-2b-SFT

Uploaded model

• Developed by: suriya7
• License: apache-2.0
• Finetuned from model : unsloth/gemma-2b

This Gemma model was trained 2x faster with Unsloth and Huggingface's TRL library.

Requirements

pip install torch
pip install transformers

Inference In Notebook

import torch

alpaca_prompt = """Below is an instruction that describes a task, paired with an input that provides further context. Write a response that appropriately completes the request.

### Instruction:
{}

### Input:
{}

### Response:
{}"""

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM
from transformers import TextStreamer

tokenizer = AutoTokenizer.from_pretrained("suriya7/Gemma-2b-SFT")
model = AutoModelForCausalLM.from_pretrained("suriya7/Gemma-2b-SFT")
device = "cuda" if torch.cuda.is_available() else "cpu"
model.to(device)
inputs = tokenizer(
[
    alpaca_prompt.format(
        "You are an AI assistant. Please ensure that the answers conclude with an end-of-sequence (EOS) token.", # instruction
        "how to cook pizza?", # input goes here
        "", # output - leave this blank for generation!
    )
], return_tensors = "pt").to(device)

text_streamer = TextStreamer(tokenizer)
_ = model.generate(**inputs, streamer = text_streamer, max_new_tokens = 250,do_sample=True,temperature=0.7,top_k=2,repetition_penalty=1.5,  # Penalize repeated responses
eos_token_id=model.config.eos_token_id)

Recommended Prompt Template

alpaca_prompt = """Below is an instruction that describes a task, paired with an input that provides further context. Write a response that appropriately completes the request.

### Instruction:
{}

### Input:
{}

### Response:
{}"""