Spaces:
Runtime error
Runtime error
File size: 9,617 Bytes
47031d7 b5d6152 47031d7 b5d6152 47031d7 02fd6bb eb5a3fb 47031d7 7b336c4 47031d7 92cdcfc b5d6152 02fd6bb eb5a3fb 47031d7 d0b5a4b 47031d7 b5d6152 |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 |
from fastapi import APIRouter, HTTPException
from fastapi.responses import StreamingResponse
from typing import Optional
import json
from time import time
import logging
from .api import InferenceApi
from .schemas import (
GenerateRequest,
EmbeddingRequest,
EmbeddingResponse,
SystemStatusResponse,
ValidationResponse,
ChatCompletionRequest,
ChatCompletionResponse, QueryExpansionResponse, QueryExpansionRequest, ChunkRerankResponse, ChunkRerankRequest
)
router = APIRouter()
logger = logging.getLogger(__name__)
api = None
def init_router(inference_api: InferenceApi):
"""Initialize router with an already setup API instance"""
global api
api = inference_api
logger.info("Router initialized with Inference API instance")
@router.post("/generate")
async def generate_text(request: GenerateRequest):
"""Generate text response from prompt"""
logger.info(f"Received generation request for prompt: {request.prompt[:50]}...")
try:
response = await api.generate_response(
prompt=request.prompt,
system_message=request.system_message,
max_new_tokens=request.max_new_tokens
)
logger.info("Successfully generated response")
return {"generated_text": response}
except Exception as e:
logger.error(f"Error in generate_text endpoint: {str(e)}")
raise HTTPException(status_code=500, detail=str(e))
@router.post("/generate/stream")
async def generate_stream(request: GenerateRequest):
"""Generate streaming text response from prompt"""
logger.info(f"Received streaming generation request for prompt: {request.prompt[:50]}...")
try:
return StreamingResponse(
api.generate_stream(
prompt=request.prompt,
system_message=request.system_message,
max_new_tokens=request.max_new_tokens
),
media_type="text/event-stream"
)
except Exception as e:
logger.error(f"Error in generate_stream endpoint: {str(e)}")
raise HTTPException(status_code=500, detail=str(e))
@router.post("/v1/chat/completions")
async def create_chat_completion(request: ChatCompletionRequest):
"""OpenAI-compatible chat completion endpoint"""
logger.info(f"Received chat completion request with {len(request.messages)} messages")
try:
# Extract the last user message, or combine messages if needed
last_message = request.messages[-1].content
if request.stream:
# For streaming, we need to create a generator that yields OpenAI-compatible chunks
async def generate_stream():
async for chunk in api.generate_stream(
prompt=last_message,
):
# Create a streaming response chunk in OpenAI format
response_chunk = {
"id": "chatcmpl-123",
"object": "chat.completion.chunk",
"created": int(time()),
"model": request.model,
"choices": [{
"index": 0,
"delta": {
"content": chunk
},
"finish_reason": None
}]
}
yield f"data: {json.dumps(response_chunk)}\n\n"
# Send the final chunk
yield f"data: [DONE]\n\n"
return StreamingResponse(
generate_stream(),
media_type="text/event-stream"
)
else:
# For non-streaming, generate the full response
response_text = await api.generate_response(
prompt=last_message,
)
# Convert to OpenAI format
return ChatCompletionResponse.from_response(
content=response_text,
model=request.model
)
except Exception as e:
logger.error(f"Error in chat completion endpoint: {str(e)}")
raise HTTPException(status_code=500, detail=str(e))
@router.post("/expand_query", response_model=QueryExpansionResponse)
async def expand_query(request: QueryExpansionRequest):
"""Expand a query for RAG processing"""
logger.info(f"Received query expansion request: {request.query[:50]}...")
try:
result = await api.expand_query(
query=request.query,
system_message=request.system_message
)
logger.info("Successfully expanded query")
return result
except FileNotFoundError as e:
logger.error(f"Template file not found: {str(e)}")
raise HTTPException(status_code=500, detail="Query expansion template not found")
except json.JSONDecodeError as e:
logger.error(f"Invalid JSON response from LLM: {str(e)}")
raise HTTPException(status_code=500, detail="Invalid response format from LLM")
except Exception as e:
logger.error(f"Error in expand_query endpoint: {str(e)}")
raise HTTPException(status_code=500, detail=str(e))
@router.post("/rerank", response_model=ChunkRerankResponse)
async def rerank_chunks(request: ChunkRerankRequest):
"""Rerank chunks based on their relevance to the query"""
logger.info(f"Received reranking request for query: {request.query[:50]}...")
try:
result = await api.rerank_chunks(
query=request.query,
chunks=request.chunks,
system_message=request.system_message
)
logger.info(f"Successfully reranked {len(request.chunks)} chunks")
return result
except Exception as e:
logger.error(f"Error in rerank_chunks endpoint: {str(e)}")
raise HTTPException(status_code=500, detail=str(e))
@router.post("/embedding", response_model=EmbeddingResponse)
async def generate_embedding(request: EmbeddingRequest):
"""Generate embedding vector from text"""
logger.info(f"Received embedding request for text: {request.text[:50]}...")
try:
embedding = await api.generate_embedding(request.text)
logger.info(f"Successfully generated embedding of dimension {len(embedding)}")
return EmbeddingResponse(
embedding=embedding,
dimension=len(embedding)
)
except Exception as e:
logger.error(f"Error in generate_embedding endpoint: {str(e)}")
raise HTTPException(status_code=500, detail=str(e))
@router.get("/system/status",
response_model=SystemStatusResponse,
summary="Check System Status",
description="Returns comprehensive system status including CPU, Memory, GPU, Storage, and Model information")
async def check_system():
"""Get system status from LLM Server"""
try:
return await api.check_system_status()
except Exception as e:
logger.error(f"Error checking system status: {str(e)}")
raise HTTPException(status_code=500, detail=str(e))
@router.get("/system/validate",
response_model=ValidationResponse,
summary="Validate System Configuration",
description="Validates system configuration, folders, and model setup")
async def validate_system():
"""Get system validation status from LLM Server"""
try:
return await api.validate_system()
except Exception as e:
logger.error(f"Error validating system: {str(e)}")
raise HTTPException(status_code=500, detail=str(e))
@router.post("/model/initialize",
summary="Initialize default or specified model",
description="Initialize model for use. Uses default model from config if none specified.")
async def initialize_model(model_name: Optional[str] = None):
"""Initialize a model for use"""
try:
return await api.initialize_model(model_name)
except Exception as e:
logger.error(f"Error initializing model: {str(e)}")
raise HTTPException(status_code=500, detail=str(e))
@router.post("/model/initialize/embedding",
summary="Initialize embedding model",
description="Initialize a separate model specifically for generating embeddings")
async def initialize_embedding_model(model_name: Optional[str] = None):
"""Initialize a model specifically for embeddings"""
try:
return await api.initialize_embedding_model(model_name)
except Exception as e:
logger.error(f"Error initializing embedding model: {str(e)}")
raise HTTPException(status_code=500, detail=str(e))
@router.post("/model/download",
summary="Download default or specified model",
description="Downloads model files. Uses default model from config if none specified.")
async def download_model(model_name: Optional[str] = None):
"""Download model files to local storage"""
try:
# Use model name from config if none provided
model_to_download = model_name or config["model"]["defaults"]["model_name"]
logger.info(f"Received request to download model: {model_to_download}")
result = await api.download_model(model_to_download)
logger.info(f"Successfully downloaded model: {model_to_download}")
return result
except Exception as e:
logger.error(f"Error downloading model: {str(e)}")
raise HTTPException(status_code=500, detail=str(e))
@router.on_event("shutdown")
async def shutdown_event():
"""Clean up resources on shutdown"""
if api:
await api.cleanup() |