Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
@@ -1,118 +1,64 @@
|
|
1 |
-
from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline, logging
|
2 |
-
from huggingface_hub import login
|
3 |
-
import torch
|
4 |
-
import os
|
5 |
import gradio as gr
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
6 |
|
7 |
-
# --- 1. Authentication (Using User-Provided Token) ---
|
8 |
-
|
9 |
-
def authenticate(token):
|
10 |
-
"""Attempts to authenticate with the provided token."""
|
11 |
-
try:
|
12 |
-
login(token=token)
|
13 |
-
return True
|
14 |
-
except Exception as e:
|
15 |
-
print(f"Authentication failed: {e}")
|
16 |
-
return False
|
17 |
-
|
18 |
-
# --- 2. Model and Tokenizer Setup ---
|
19 |
-
|
20 |
-
def load_model_and_tokenizer(model_name="google/gemma-3-1b-it"):
|
21 |
-
"""Loads the model and tokenizer."""
|
22 |
-
try:
|
23 |
-
logging.set_verbosity_error()
|
24 |
-
tokenizer = AutoTokenizer.from_pretrained(model_name)
|
25 |
-
model = AutoModelForCausalLM.from_pretrained(
|
26 |
-
model_name,
|
27 |
-
device_map="auto",
|
28 |
-
torch_dtype=torch.bfloat16,
|
29 |
-
attn_implementation="flash_attention_2"
|
30 |
-
)
|
31 |
-
return model, tokenizer
|
32 |
-
except Exception as e:
|
33 |
-
print(f"ERROR: Failed to load model/tokenizer: {e}")
|
34 |
-
raise # Re-raise for Gradio
|
35 |
-
|
36 |
-
# --- 3. Chat Template Function ---
|
37 |
-
|
38 |
-
def apply_chat_template(messages, tokenizer):
|
39 |
-
"""Applies the chat template."""
|
40 |
-
try:
|
41 |
-
if hasattr(tokenizer, "chat_template") and tokenizer.chat_template:
|
42 |
-
return tokenizer.apply_chat_template(
|
43 |
-
messages, tokenize=False, add_generation_prompt=True
|
44 |
-
)
|
45 |
-
else:
|
46 |
-
print("WARNING: Tokenizer lacks chat_template. Using fallback.")
|
47 |
-
chat_template = "{% for message in messages %}" \
|
48 |
-
"{{ '<start_of_turn>' + message['role'] + '\n' + message['content'] + '<end_of_turn>\n' }}" \
|
49 |
-
"{% endfor %}" \
|
50 |
-
"{% if add_generation_prompt %}{{ '<start_of_turn>model\n' }}{% endif %}"
|
51 |
-
return tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True, chat_template=chat_template)
|
52 |
-
except Exception as e:
|
53 |
-
print(f"ERROR: Chat template application failed: {e}")
|
54 |
-
raise
|
55 |
-
|
56 |
-
# --- 4. Text Generation Function ---
|
57 |
-
|
58 |
-
def generate_response(messages, model, tokenizer, max_new_tokens=256, temperature=0.7, top_k=50, top_p=0.95, repetition_penalty=1.2):
|
59 |
-
"""Generates a response."""
|
60 |
-
prompt = apply_chat_template(messages, tokenizer)
|
61 |
-
try:
|
62 |
-
pipeline_instance = pipeline(
|
63 |
-
"text-generation", model=model, tokenizer=tokenizer,
|
64 |
-
torch_dtype=torch.bfloat16, device_map="auto",
|
65 |
-
model_kwargs={"attn_implementation": "flash_attention_2"}
|
66 |
-
)
|
67 |
-
outputs = pipeline_instance(
|
68 |
-
prompt, max_new_tokens=max_new_tokens, do_sample=True,
|
69 |
-
temperature=temperature, top_k=top_k, top_p=top_p,
|
70 |
-
repetition_penalty=repetition_penalty, pad_token_id=tokenizer.eos_token_id
|
71 |
-
)
|
72 |
-
return outputs[0]["generated_text"][len(prompt):].strip()
|
73 |
-
except Exception as e:
|
74 |
-
print(f"ERROR: Response generation failed: {e}")
|
75 |
-
raise
|
76 |
-
|
77 |
-
# --- 5. Gradio Interface ---
|
78 |
-
model = None # Initialize model and tokenizer as global variables
|
79 |
-
tokenizer = None
|
80 |
-
|
81 |
-
def chat(token, message, history):
|
82 |
-
global model, tokenizer # Access the global model and tokenizer
|
83 |
-
|
84 |
-
if not authenticate(token):
|
85 |
-
return "Authentication failed. Please enter a valid Hugging Face token.", history
|
86 |
-
|
87 |
-
if model is None or tokenizer is None:
|
88 |
-
try:
|
89 |
-
model, tokenizer = load_model_and_tokenizer()
|
90 |
-
except Exception as e:
|
91 |
-
return f"Model loading error: {e}", history
|
92 |
-
|
93 |
-
if not history:
|
94 |
-
history = []
|
95 |
-
messages = [{"role": "user", "content": msg} for msg, _ in history]
|
96 |
-
messages.extend([{"role": "model", "content": resp} for _, resp in history if resp])
|
97 |
messages.append({"role": "user", "content": message})
|
98 |
|
99 |
-
|
100 |
-
|
101 |
-
|
102 |
-
|
103 |
-
|
104 |
-
|
105 |
-
|
106 |
-
|
107 |
-
|
108 |
-
|
109 |
-
|
110 |
-
|
111 |
-
|
112 |
-
|
113 |
-
|
114 |
-
|
115 |
-
|
116 |
-
|
117 |
-
|
118 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
import gradio as gr
|
2 |
+
from huggingface_hub import InferenceClient
|
3 |
+
|
4 |
+
"""
|
5 |
+
For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
|
6 |
+
"""
|
7 |
+
client = InferenceClient("google/gemma-3-1b-it")
|
8 |
+
|
9 |
+
|
10 |
+
def respond(
|
11 |
+
message,
|
12 |
+
history: list[tuple[str, str]],
|
13 |
+
system_message,
|
14 |
+
max_tokens,
|
15 |
+
temperature,
|
16 |
+
top_p,
|
17 |
+
):
|
18 |
+
messages = [{"role": "system", "content": system_message}]
|
19 |
+
|
20 |
+
for val in history:
|
21 |
+
if val[0]:
|
22 |
+
messages.append({"role": "user", "content": val[0]})
|
23 |
+
if val[1]:
|
24 |
+
messages.append({"role": "assistant", "content": val[1]})
|
25 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
26 |
messages.append({"role": "user", "content": message})
|
27 |
|
28 |
+
response = ""
|
29 |
+
|
30 |
+
for message in client.chat_completion(
|
31 |
+
messages,
|
32 |
+
max_tokens=max_tokens,
|
33 |
+
stream=True,
|
34 |
+
temperature=temperature,
|
35 |
+
top_p=top_p,
|
36 |
+
):
|
37 |
+
token = message.choices[0].delta.content
|
38 |
+
|
39 |
+
response += token
|
40 |
+
yield response
|
41 |
+
|
42 |
+
|
43 |
+
"""
|
44 |
+
For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
|
45 |
+
"""
|
46 |
+
demo = gr.ChatInterface(
|
47 |
+
respond,
|
48 |
+
additional_inputs=[
|
49 |
+
gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
|
50 |
+
gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
|
51 |
+
gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
|
52 |
+
gr.Slider(
|
53 |
+
minimum=0.1,
|
54 |
+
maximum=1.0,
|
55 |
+
value=0.95,
|
56 |
+
step=0.05,
|
57 |
+
label="Top-p (nucleus sampling)",
|
58 |
+
),
|
59 |
+
],
|
60 |
+
)
|
61 |
+
|
62 |
+
|
63 |
+
if __name__ == "__main__":
|
64 |
+
demo.launch()
|