Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
@@ -4,133 +4,115 @@ import torch
|
|
4 |
import os
|
5 |
import gradio as gr
|
6 |
|
7 |
-
# --- 1. Authentication (Using
|
8 |
|
9 |
-
|
10 |
-
|
11 |
-
|
12 |
-
|
13 |
-
|
14 |
-
|
15 |
-
|
16 |
-
|
17 |
-
# The login() call below will now automatically use the environment variable.
|
18 |
-
login()
|
19 |
|
20 |
-
# --- 2. Model and Tokenizer Setup
|
21 |
|
22 |
def load_model_and_tokenizer(model_name="google/gemma-3-1b-it"):
|
23 |
-
"""Loads the model and tokenizer
|
24 |
try:
|
25 |
-
# Suppress unnecessary warning messages from transformers
|
26 |
logging.set_verbosity_error()
|
27 |
-
|
28 |
tokenizer = AutoTokenizer.from_pretrained(model_name)
|
29 |
model = AutoModelForCausalLM.from_pretrained(
|
30 |
model_name,
|
31 |
-
device_map="auto",
|
32 |
-
torch_dtype=torch.bfloat16,
|
33 |
-
attn_implementation="flash_attention_2"
|
34 |
)
|
35 |
return model, tokenizer
|
36 |
-
|
37 |
except Exception as e:
|
38 |
-
print(f"ERROR: Failed to load model
|
39 |
-
|
40 |
-
print("1. Ensure you have a Hugging Face account and have accepted the model's terms.")
|
41 |
-
print("2. Verify your internet connection.")
|
42 |
-
print("3. Double-check the model name: 'google/gemma-3-1b-it'")
|
43 |
-
print("4. Ensure you are properly authenticated using a Repository Secret (see above).")
|
44 |
-
print("5. If using a GPU, ensure your CUDA drivers and PyTorch are correctly installed.")
|
45 |
-
# Instead of exiting, raise the exception to be caught by Gradio
|
46 |
-
raise
|
47 |
-
|
48 |
-
model, tokenizer = load_model_and_tokenizer()
|
49 |
-
|
50 |
|
51 |
-
# --- 3. Chat Template Function
|
52 |
|
53 |
def apply_chat_template(messages, tokenizer):
|
54 |
-
"""Applies the
|
55 |
try:
|
56 |
if hasattr(tokenizer, "chat_template") and tokenizer.chat_template:
|
57 |
return tokenizer.apply_chat_template(
|
58 |
messages, tokenize=False, add_generation_prompt=True
|
59 |
)
|
60 |
else:
|
61 |
-
print("WARNING: Tokenizer
|
62 |
chat_template = "{% for message in messages %}" \
|
63 |
"{{ '<start_of_turn>' + message['role'] + '\n' + message['content'] + '<end_of_turn>\n' }}" \
|
64 |
"{% endfor %}" \
|
65 |
"{% if add_generation_prompt %}{{ '<start_of_turn>model\n' }}{% endif %}"
|
66 |
return tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True, chat_template=chat_template)
|
67 |
-
|
68 |
except Exception as e:
|
69 |
-
print(f"ERROR:
|
70 |
-
raise
|
71 |
-
|
72 |
|
73 |
# --- 4. Text Generation Function ---
|
74 |
|
75 |
def generate_response(messages, model, tokenizer, max_new_tokens=256, temperature=0.7, top_k=50, top_p=0.95, repetition_penalty=1.2):
|
76 |
"""Generates a response."""
|
77 |
prompt = apply_chat_template(messages, tokenizer)
|
78 |
-
|
79 |
try:
|
80 |
pipeline_instance = pipeline(
|
81 |
-
"text-generation",
|
82 |
-
|
83 |
-
tokenizer=tokenizer,
|
84 |
-
torch_dtype=torch.bfloat16,
|
85 |
-
device_map="auto",
|
86 |
model_kwargs={"attn_implementation": "flash_attention_2"}
|
87 |
-
|
88 |
-
|
89 |
outputs = pipeline_instance(
|
90 |
-
prompt,
|
91 |
-
|
92 |
-
|
93 |
-
temperature=temperature,
|
94 |
-
top_k=top_k,
|
95 |
-
top_p=top_p,
|
96 |
-
repetition_penalty=repetition_penalty,
|
97 |
-
pad_token_id=tokenizer.eos_token_id,
|
98 |
)
|
|
|
|
|
|
|
|
|
99 |
|
100 |
-
|
101 |
-
|
|
|
102 |
|
103 |
-
|
104 |
-
|
105 |
-
raise # Re-raise the exception
|
106 |
|
|
|
|
|
107 |
|
108 |
-
|
|
|
|
|
|
|
|
|
109 |
|
110 |
-
def predict(message, history):
|
111 |
if not history:
|
112 |
history = []
|
113 |
-
messages = []
|
114 |
-
for
|
115 |
-
messages.append({"role": "user", "content": user_msg})
|
116 |
-
if bot_response: # Check if bot_response is not None
|
117 |
-
messages.append({"role": "model", "content": bot_response})
|
118 |
messages.append({"role": "user", "content": message})
|
119 |
|
120 |
try:
|
121 |
-
|
122 |
-
|
123 |
-
|
124 |
except Exception as e:
|
125 |
-
|
126 |
-
return f"Error: {e}", history
|
127 |
-
|
128 |
|
129 |
with gr.Blocks() as demo:
|
130 |
-
|
131 |
-
|
132 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
133 |
|
134 |
-
msg.submit(predict, [msg, chatbot], [msg, chatbot])
|
135 |
|
136 |
demo.launch()
|
|
|
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 |
try:
|
100 |
+
response = generate_response(messages, model, tokenizer)
|
101 |
+
history.append((message, response))
|
102 |
+
return "", history
|
103 |
except Exception as e:
|
104 |
+
return f"Error during generation: {e}", history
|
|
|
|
|
105 |
|
106 |
with gr.Blocks() as demo:
|
107 |
+
gr.Markdown("# Gemma Chatbot")
|
108 |
+
gr.Markdown("Enter your Hugging Face API token (read access required):")
|
109 |
+
token_input = gr.Textbox(label="Hugging Face Token", type="password") # Use type="password"
|
110 |
+
chatbot = gr.Chatbot(label="Chat", height=400)
|
111 |
+
msg_input = gr.Textbox(label="Message", placeholder="Ask me anything!")
|
112 |
+
clear_btn = gr.ClearButton([msg_input, chatbot])
|
113 |
+
|
114 |
+
msg_input.submit(chat, [token_input, msg_input, chatbot], [msg_input, chatbot])
|
115 |
+
clear_btn.click(lambda: (None, []), [], [msg_input, chatbot])
|
116 |
|
|
|
117 |
|
118 |
demo.launch()
|