app.py

# import torch
from peft import PeftModel
import torch
import transformers
import gradio as gr

assert (
    "LlamaTokenizer" in transformers._import_structure["models.llama"]
), "LLaMA is now in HuggingFace's main branch.\nPlease reinstall it: pip uninstall transformers && pip install git+https://github.com/huggingface/transformers.git"
from transformers import LlamaTokenizer, LlamaForCausalLM, GenerationConfig

tokenizer = LlamaTokenizer.from_pretrained("decapoda-research/llama-7b-hf")

BASE_MODEL = "decapoda-research/llama-7b-hf"
LORA_WEIGHTS = "OdiaGenAI/odiagenAI-model-v0"

if torch.cuda.is_available():
    device = "cuda"
else:
    device = "cpu"

try:
    if torch.backends.mps.is_available():
        device = "mps"
except:
    pass

if device == "cuda":
    model = LlamaForCausalLM.from_pretrained(
        BASE_MODEL,
        load_in_8bit=False,
        torch_dtype=torch.float16,
        device_map="auto",
    )
    model = PeftModel.from_pretrained(
        model, LORA_WEIGHTS, torch_dtype=torch.float16, force_download=True
    )
elif device == "mps":
    model = LlamaForCausalLM.from_pretrained(
        BASE_MODEL,
        device_map={"": device},
        torch_dtype=torch.float16,
    )
    model = PeftModel.from_pretrained(
        model,
        LORA_WEIGHTS,
        device_map={"": device},
        torch_dtype=torch.float16,
    )
else:
    model = LlamaForCausalLM.from_pretrained(
        BASE_MODEL, device_map={"": device}, low_cpu_mem_usage=True
    )
    model = PeftModel.from_pretrained(
        model,
        LORA_WEIGHTS,
        device_map={"": device},
    )



def generate_prompt(instruction, input=None):
    if input:
        return f"""ନିମ୍ନରେ ଏକ ନିର୍ଦ୍ଦେଶନାମା ଯାହାକି ଏକ କାର୍ଯ୍ୟକୁ ବର୍ଣ୍ଣନା କରେ, ଏକ ଇନପୁଟ୍ ସହିତ ଯୋଡି ଯାହା ପରବର୍ତ୍ତୀ ପ୍ରସଙ୍ଗ ପ୍ରଦାନ କରେ | ଏକ ପ୍ରତିକ୍ରିୟା ଲେଖନ୍ତୁ ଯାହା ଅନୁରୋଧକୁ ସଠିକ୍ ଭାବରେ ସମାପ୍ତ କରେ |
### ନିର୍ଦ୍ଦେଶ:
{instruction}
### ଇନପୁଟ୍:
{input}
### ପ୍ରତିକ୍ରିୟା:"""
    else:
        return f"""ନିମ୍ନରେ ଏକ ନିର୍ଦ୍ଦେଶ ଯାହାକି ଏକ କାର୍ଯ୍ୟକୁ ବର୍ଣ୍ଣନା କରେ | ଏକ ପ୍ରତିକ୍ରିୟା ଲେଖନ୍ତୁ ଯାହା ଅନୁରୋଧକୁ ସଠିକ୍ ଭାବରେ ସମାପ୍ତ କରେ |
### ନିର୍ଦ୍ଦେଶ:
{instruction}
### ପ୍ରତିକ୍ରିୟା:"""




if device != "cpu":
    model.half()
model.eval()
if torch.__version__ >= "2":
    model = torch.compile(model)


def evaluate(
    instruction,
    input=None,
    temperature=0.1,
    top_p=0.75,
    top_k=40,
    num_beams=4,
    max_new_tokens=128,
    **kwargs,
):
    prompt = generate_prompt(instruction, input)
    print(prompt)
    inputs = tokenizer(prompt, return_tensors="pt")
    print(inputs)
    input_ids = inputs["input_ids"].to(device)
    print(input_ids)
    generation_config = GenerationConfig(
        temperature=temperature,
        top_p=top_p,
        top_k=top_k,
        num_beams=num_beams,
        **kwargs,
    )
    with torch.no_grad():
        generation_output = model.generate(
            input_ids=input_ids,
            generation_config=generation_config,
            return_dict_in_generate=True,
            output_scores=True,
            max_new_tokens=max_new_tokens,
        )
        print(generation_output)
    s = generation_output.sequences[0]
    print(s)
    output = tokenizer.decode(s)
    print(output)
    return output.split("### ପ୍ରତିକ୍ରିୟା:")[1].strip()


g = gr.Interface(
    fn=evaluate,
    inputs=[
        gr.components.Textbox(
            lines=2, label="Instruction", placeholder="Tell me about alpacas."
        ),
        gr.components.Textbox(lines=2, label="Input", placeholder="none"),
        gr.components.Slider(minimum=0, maximum=1, value=0.1, label="Temperature"),
        gr.components.Slider(minimum=0, maximum=1, value=0.75, label="Top p"),
        gr.components.Slider(minimum=0, maximum=100, step=1, value=40, label="Top k"),
        gr.components.Slider(minimum=1, maximum=4, step=1, value=4, label="Beams"),
        gr.components.Slider(
            minimum=1, maximum=512, step=1, value=128, label="Max tokens"
        ),
    ],
    outputs=[
        gr.inputs.Textbox(
            lines=5,
            label="Output",
        )
    ],
    title="🦙🌲 OdiaGenAI-LoRA",
    description="OdiaGenAI-LoRA is a 7B-parameter LLaMA model fine-tuned to follow Odia instructions. It is trained on the [Stanford Alpaca](https://github.com/tatsu-lab/stanford_alpaca) Odia translated dataset and makes use of the Huggingface LLaMA implementation. For more information, please visit [the project's website](https://github.com/shantipriyap/OdiaGenAI).",
)
g.queue(concurrency_count=1)
g.launch()