File size: 2,267 Bytes
021c2c9 afa7e4f 021c2c9 afa7e4f 021c2c9 afa7e4f 021c2c9 afa7e4f 021c2c9 afa7e4f 021c2c9 afa7e4f 021c2c9 afa7e4f 021c2c9 afa7e4f 021c2c9 afa7e4f 021c2c9 afa7e4f 021c2c9 afa7e4f 021c2c9 afa7e4f 021c2c9 afa7e4f 021c2c9 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 |
import torch
import transformers
import gradio as gr
from unsloth import FastLanguageModel
# Load the fine-tuned Unsloth model
max_seq_length = 2048 # Adjust based on your training
dtype = None # None for auto detection
def load_model():
model, tokenizer = FastLanguageModel.from_pretrained(
model_name="ivwhy/lora_model", # Your fine-tuned model path
max_seq_length=max_seq_length,
dtype=dtype,
load_in_4bit=True # Optional: load in 4-bit for efficiency
)
# Optional: Add special tokens for chat if needed
tokenizer.pad_token = tokenizer.eos_token
# Create the pipeline
pipeline = transformers.pipeline(
"text-generation",
model=model,
tokenizer=tokenizer,
device=0 if torch.cuda.is_available() else -1 # Use GPU if available
)
return pipeline, tokenizer
# Load model globally
generation_pipeline, tokenizer = load_model()
def chat_function(message, history, system_prompt, max_new_tokens, temperature):
messages = [
{"role": "system", "content": system_prompt},
{"role": "user", "content": message}
]
# Apply chat template
prompt = tokenizer.apply_chat_template(
messages,
tokenize=False,
add_generation_prompt=True,
)
# Define terminators
terminators = [
tokenizer.eos_token_id,
tokenizer.convert_tokens_to_ids("<|eot_id|>")
]
# Generate response
outputs = generation_pipeline(
prompt,
max_new_tokens=max_new_tokens,
eos_token_id=terminators,
do_sample=True,
temperature=temperature,
top_p=0.9,
)
# Extract and return just the generated text
return outputs[0]["generated_text"][len(prompt):]
# Create Gradio interface
demo = gr.ChatInterface(
chat_function,
textbox=gr.Textbox(placeholder="Enter message here", container=False, scale=7),
chatbot=gr.Chatbot(height=400),
additional_inputs=[
gr.Textbox("You are helpful AI", label="System Prompt"),
gr.Slider(minimum=1, maximum=4000, value=500, label="Max New Tokens"),
gr.Slider(minimum=0, maximum=1, value=0.7, label="Temperature")
]
)
if __name__ == "__main__":
demo.launch() |