File size: 7,843 Bytes
2dc4fb9
 
a971d1c
c51f8c4
e4a835d
3683c2c
 
 
f817cfc
4b20d59
a971d1c
c51f8c4
7b37201
e28d45d
 
 
2dc4fb9
e4a835d
02d77c2
f817cfc
 
2dc4fb9
02d77c2
2dc4fb9
7ac8cfa
c2c3825
02d77c2
e4a835d
990db9b
 
 
 
e4a835d
2dc4fb9
 
4b20d59
2dc4fb9
7b37201
c51f8c4
 
e28d45d
 
 
c51f8c4
 
 
 
 
02d77c2
c51f8c4
 
 
 
 
 
 
 
 
 
 
 
 
 
02d77c2
 
c51f8c4
 
 
02d77c2
 
c51f8c4
 
 
 
 
e28d45d
 
 
4b20d59
 
e28d45d
3683c2c
a971d1c
e28d45d
 
c2c3825
8dc383f
e28d45d
 
8dc383f
a971d1c
8dc383f
02d77c2
4b20d59
c51f8c4
e28d45d
a971d1c
c2c3825
c51f8c4
e28d45d
 
c51f8c4
 
02d77c2
c51f8c4
4b20d59
02d77c2
8dc383f
a971d1c
 
 
c2c3825
02d77c2
a971d1c
 
02d77c2
 
e28d45d
a971d1c
8dc383f
e4a835d
e28d45d
02d77c2
e28d45d
8dc383f
 
4b20d59
a971d1c
8dc383f
e28d45d
 
 
 
 
2dc4fb9
e28d45d
 
 
7ac8cfa
e28d45d
 
 
02d77c2
e28d45d
 
c2c3825
e28d45d
 
 
 
 
 
 
 
 
 
 
 
 
02d77c2
e28d45d
 
 
02d77c2
e28d45d
 
02d77c2
e28d45d
 
 
 
 
 
 
 
 
 
02d77c2
e28d45d
 
 
427e302
e28d45d
 
02d77c2
e28d45d
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3683c2c
e28d45d
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2dc4fb9
 
e28d45d
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
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
import gradio as gr
from faster_whisper import WhisperModel
from transformers import AutoTokenizer, AutoModelForCausalLM
import torch
import requests
import base64
import tempfile
import os
import logging
import time
from datetime import datetime
from concurrent.futures import ThreadPoolExecutor
from html.parser import HTMLParser
from fastapi import FastAPI, Request
from fastapi.responses import JSONResponse
import uvicorn

# Setup logging
logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(message)s')
logger = logging.getLogger(__name__)

# Initialize models
logger.info("Loading Whisper-tiny...")
whisper_model = WhisperModel("tiny", device="cpu", compute_type="int8")

logger.info("Loading SmolLM2-360M-Instruct...")
model_name = "HuggingFaceTB/SmolLM2-360M-Instruct"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(
    model_name,
    torch_dtype=torch.float32,
    device_map="cpu",
    low_cpu_mem_usage=True
)

logger.info("All models loaded!")

TAVILY_API_KEY = os.getenv('TAVILY_API_KEY', '')
BRAVE_API_KEY = os.getenv('BRAVE_API_KEY', '')

def search_parallel(query):
    """Simplified search - just DuckDuckGo for speed"""
    logger.info("[SEARCH] Starting...")
    try:
        response = requests.get(
            'https://html.duckduckgo.com/html/',
            params={'q': query},
            headers={'User-Agent': 'Mozilla/5.0'},
            timeout=1.5
        )
        if response.status_code == 200:
            class DDGParser(HTMLParser):
                def __init__(self):
                    super().__init__()
                    self.results = []
                    self.in_result = False
                    self.current_text = ""
                
                def handle_starttag(self, tag, attrs):
                    if tag == 'a' and any(k == 'class' and 'result__a' in v for k, v in attrs):
                        self.in_result = True
                
                def handle_data(self, data):
                    if self.in_result and data.strip():
                        self.current_text += data.strip() + " "
                
                def handle_endtag(self, tag):
                    if tag == 'a' and self.in_result:
                        if self.current_text:
                            self.results.append(self.current_text.strip()[:120])
                        self.current_text = ""
                        self.in_result = False
            
            parser = DDGParser()
            parser.feed(response.text)
            result = "\n".join([f"• {r}" for r in parser.results[:2]]) if parser.results else "No results"
            logger.info("[SEARCH] ✓")
            return result, "DuckDuckGo"
    except:
        pass
    return "No search results", "None"

def generate_answer(text_input):
    """Main answer generation"""
    logger.info(f"[AI] Question: {text_input[:60]}...")
    
    try:
        if not text_input or not text_input.strip():
            return "No input provided"
        
        current_date = datetime.now().strftime("%B %d, %Y")
        
        # Search
        search_start = time.time()
        search_results, search_engine = search_parallel(text_input)
        logger.info(f"[AI] Search: {time.time()-search_start:.2f}s")
        
        # Generate
        messages = [
            {"role": "system", "content": f"Today is {current_date}. Answer briefly using search results (60-80 words)."},
            {"role": "user", "content": f"Search:\n{search_results}\n\nQ: {text_input}\nA:"}
        ]
        
        prompt = f"<|im_start|>system\n{messages[0]['content']}<|im_end|>\n<|im_start|>user\n{messages[1]['content']}<|im_end|>\n<|im_start|>assistant\n"
        
        gen_start = time.time()
        inputs = tokenizer(prompt, return_tensors="pt", truncation=True, max_length=800)
        
        with torch.no_grad():
            outputs = model.generate(
                **inputs,
                max_new_tokens=80,
                temperature=0.7,
                do_sample=True,
                top_p=0.9,
                top_k=40,
                repetition_penalty=1.15,
                pad_token_id=tokenizer.eos_token_id
            )
        
        answer = tokenizer.decode(outputs[0][inputs['input_ids'].shape[1]:], skip_special_tokens=True).strip()
        logger.info(f"[AI] Gen: {time.time()-gen_start:.2f}s | ✓")
        
        return f"{answer}\n\n**Source:** {search_engine}"
        
    except Exception as e:
        logger.error(f"[AI] Error: {str(e)}")
        return f"Error: {str(e)}"

def transcribe_audio_base64(audio_base64):
    """Transcribe audio"""
    logger.info("[STT] Start")
    try:
        audio_bytes = base64.b64decode(audio_base64)
        
        with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as temp_audio:
            temp_audio.write(audio_bytes)
            temp_path = temp_audio.name
        
        segments, _ = whisper_model.transcribe(temp_path, language="en", beam_size=1)
        transcription = " ".join([seg.text for seg in segments])
        os.unlink(temp_path)
        
        logger.info("[STT] ✓")
        return transcription.strip()
        
    except Exception as e:
        logger.error(f"[STT] Error: {str(e)}")
        return ""

# Create FastAPI app for Pluely endpoints
app = FastAPI()

@app.post("/api/stt")
async def api_stt(request: Request):
    """Direct STT endpoint for Pluely"""
    try:
        body = await request.json()
        logger.info(f"[API STT] Received: {body}")
        
        audio_base64 = body.get("audio", "")
        if not audio_base64:
            return JSONResponse({"error": "No audio data"}, status_code=400)
        
        text = transcribe_audio_base64(audio_base64)
        return JSONResponse({"text": text})
        
    except Exception as e:
        logger.error(f"[API STT] Error: {str(e)}")
        return JSONResponse({"error": str(e)}, status_code=500)

@app.post("/api/ai")
async def api_ai(request: Request):
    """Direct AI endpoint for Pluely"""
    try:
        body = await request.json()
        logger.info(f"[API AI] Received: {body}")
        
        question = body.get("text", "")
        if not question:
            return JSONResponse({"error": "No text provided"}, status_code=400)
        
        answer = generate_answer(question)
        return JSONResponse({"answer": answer})
        
    except Exception as e:
        logger.error(f"[API AI] Error: {str(e)}")
        return JSONResponse({"error": str(e)}, status_code=500)

@app.get("/health")
async def health():
    """Health check"""
    return {"status": "ok", "model": "SmolLM2-360M"}

# Gradio UI (optional, for testing)
with gr.Blocks(title="Fast Q&A", theme=gr.themes.Soft()) as demo:
    gr.Markdown("""
    # ⚡ Ultra-Fast Q&A System
    **SmolLM2-360M** + **Direct REST API** for Pluely
    
    ## Pluely Configuration:
    
    ### STT Endpoint:
    ```
    curl -X POST https://archcoder-basic-app.hf.space/api/stt -H "Content-Type: application/json" -d '{"audio": "{{AUDIO_BASE64}}"}'
    ```
    **Response Path:** `text`
    
    ### AI Endpoint:
    ```
    curl -X POST https://archcoder-basic-app.hf.space/api/ai -H "Content-Type: application/json" -d '{"text": "{{TEXT}}"}'
    ```
    **Response Path:** `answer`
    """)
    
    with gr.Tab("Test"):
        with gr.Row():
            test_input = gr.Textbox(label="Question", placeholder="Ask anything...")
            test_btn = gr.Button("🚀 Test")
        test_output = gr.Textbox(label="Answer", lines=8)
        
        test_btn.click(fn=generate_answer, inputs=[test_input], outputs=[test_output])

# Mount Gradio to FastAPI
app = gr.mount_gradio_app(app, demo, path="/")

if __name__ == "__main__":
    uvicorn.run(app, host="0.0.0.0", port=7860)