Update app.py
Browse files
app.py
CHANGED
@@ -1,17 +1,12 @@
|
|
1 |
from fastapi import FastAPI, Query, HTTPException
|
2 |
-
import os
|
3 |
-
os.environ["CUDA_VISIBLE_DEVICES"] = ""
|
4 |
import torch
|
5 |
import re
|
6 |
import time
|
7 |
import logging
|
8 |
import os
|
9 |
-
import gc
|
10 |
-
import json
|
11 |
from transformers import AutoTokenizer, GenerationConfig
|
12 |
from peft import AutoPeftModelForCausalLM
|
13 |
-
|
14 |
-
from unsloth import FastLanguageModel
|
15 |
|
16 |
# -------- CONFIGURAÇÕES DE OTIMIZAÇÃO --------
|
17 |
os.environ["TOKENIZERS_PARALLELISM"] = "false"
|
@@ -20,16 +15,19 @@ os.environ["MKL_NUM_THREADS"] = "2"
|
|
20 |
torch.set_num_threads(2)
|
21 |
torch.set_num_interop_threads(1)
|
22 |
|
23 |
-
# -------- LOGGING --------
|
24 |
logging.basicConfig(level=logging.INFO, format="%(asctime)s [%(levelname)s] %(message)s")
|
25 |
log = logging.getLogger("news-filter")
|
26 |
|
27 |
-
# --------
|
28 |
model_name = "habulaj/filterinstruct180"
|
29 |
log.info("🚀 Carregando modelo e tokenizer...")
|
30 |
|
31 |
-
tokenizer = AutoTokenizer.from_pretrained(
|
32 |
-
|
|
|
|
|
|
|
33 |
|
34 |
if tokenizer.pad_token is None:
|
35 |
tokenizer.pad_token = tokenizer.eos_token
|
@@ -39,11 +37,12 @@ model = AutoPeftModelForCausalLM.from_pretrained(
|
|
39 |
device_map="cpu",
|
40 |
torch_dtype=torch.bfloat16,
|
41 |
low_cpu_mem_usage=True,
|
|
|
42 |
trust_remote_code=True
|
43 |
)
|
44 |
-
|
45 |
model.eval()
|
46 |
-
log.info("✅ Modelo carregado (
|
47 |
|
48 |
generation_config = GenerationConfig(
|
49 |
max_new_tokens=128,
|
@@ -53,113 +52,120 @@ generation_config = GenerationConfig(
|
|
53 |
use_cache=True,
|
54 |
eos_token_id=tokenizer.eos_token_id,
|
55 |
pad_token_id=tokenizer.eos_token_id,
|
|
|
56 |
repetition_penalty=1.1,
|
57 |
length_penalty=1.0
|
58 |
)
|
59 |
|
60 |
-
# -------- FASTAPI --------
|
61 |
app = FastAPI(title="News Filter JSON API")
|
62 |
|
63 |
@app.get("/")
|
64 |
def read_root():
|
65 |
return {"message": "News Filter JSON API is running!", "docs": "/docs"}
|
66 |
|
67 |
-
@app.get("/filter")
|
68 |
-
def get_filter(
|
69 |
-
title: str = Query(..., description="News title"),
|
70 |
-
content: str = Query(..., description="News content")
|
71 |
-
):
|
72 |
-
try:
|
73 |
-
result = infer_filter(title, content)
|
74 |
-
try:
|
75 |
-
return {"result": json.loads(result)}
|
76 |
-
except json.JSONDecodeError:
|
77 |
-
return {"result": result, "warning": "Returned as string due to JSON parsing error"}
|
78 |
-
except HTTPException as he:
|
79 |
-
raise he
|
80 |
-
except Exception as e:
|
81 |
-
log.exception("❌ Erro inesperado:")
|
82 |
-
raise HTTPException(status_code=500, detail="Internal server error during inference.")
|
83 |
-
|
84 |
-
@app.on_event("startup")
|
85 |
-
async def warmup():
|
86 |
-
log.info("🔥 Executando warmup...")
|
87 |
-
try:
|
88 |
-
infer_filter("Test title", "Test content")
|
89 |
-
log.info("✅ Warmup concluído.")
|
90 |
-
except Exception as e:
|
91 |
-
log.warning(f"⚠️ Warmup falhou: {e}")
|
92 |
-
|
93 |
# -------- INFERÊNCIA --------
|
94 |
def infer_filter(title, content):
|
95 |
-
messages = [
|
96 |
-
{
|
97 |
-
"role": "user",
|
98 |
-
"content": """Analyze the news title and content, and return the filters in JSON format with the defined fields.
|
99 |
-
|
100 |
-
Please respond ONLY with the JSON filter, do NOT add any explanations, system messages, or extra text.
|
101 |
-
|
102 |
-
Title: "New 'Star Wars' Movie Announced"
|
103 |
-
Content: "Lucasfilm confirmed a new Star Wars movie set to release in 2026, directed by a rising filmmaker."
|
104 |
-
"""
|
105 |
-
},
|
106 |
-
{
|
107 |
-
"role": "assistant",
|
108 |
-
"content": '{ "death_related": false, "relevance": "high", "global_interest": true, "entity_type": "movie", "entity_name": "Star Wars", "breaking_news": true, "has_video_content": false }'
|
109 |
-
},
|
110 |
-
{
|
111 |
-
"role": "user",
|
112 |
-
"content": """Analyze the news title and content, and return the filters in JSON format with the defined fields.
|
113 |
-
|
114 |
-
Please respond ONLY with the JSON filter, do NOT add any explanations, system messages, or extra text.
|
115 |
-
|
116 |
-
Title: "Legendary Musician Carlos Mendes Dies at 78"
|
117 |
-
Content: "Carlos Mendes, the internationally acclaimed Brazilian guitarist and composer known for blending traditional bossa nova with modern jazz, has died at the age of 78."
|
118 |
-
"""
|
119 |
-
},
|
120 |
-
{
|
121 |
-
"role": "assistant",
|
122 |
-
"content": '{ "death_related": true, "relevance": "high", "global_interest": true, "entity_type": "person", "entity_name": "Carlos Mendes", "breaking_news": true, "has_video_content": false }'
|
123 |
-
},
|
124 |
-
{
|
125 |
-
"role": "user",
|
126 |
-
"content": f"""Analyze the news title and content, and return the filters in JSON format with the defined fields.
|
127 |
-
|
128 |
-
Please respond ONLY with the JSON filter, do NOT add any explanations, system messages, or extra text.
|
129 |
-
|
130 |
-
Title: "{title}"
|
131 |
-
Content: "{content}"
|
132 |
-
"""
|
133 |
-
}
|
134 |
-
]
|
135 |
-
|
136 |
log.info(f"🧠 Inferência iniciada para: {title}")
|
137 |
start_time = time.time()
|
138 |
|
139 |
-
|
140 |
-
|
141 |
-
|
142 |
-
|
143 |
return_tensors="pt",
|
144 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
145 |
|
146 |
with torch.no_grad(), torch.inference_mode():
|
147 |
outputs = model.generate(
|
148 |
-
input_ids=
|
|
|
149 |
generation_config=generation_config,
|
|
|
|
|
|
|
150 |
)
|
151 |
|
152 |
-
|
153 |
-
|
154 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
155 |
|
156 |
-
json_str = extract_json(generated)
|
157 |
duration = time.time() - start_time
|
158 |
log.info(f"✅ JSON extraído em {duration:.2f}s")
|
159 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
160 |
|
161 |
def extract_json(text):
|
162 |
-
match = re.search(r'\{
|
163 |
if match:
|
164 |
-
|
165 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
from fastapi import FastAPI, Query, HTTPException
|
|
|
|
|
2 |
import torch
|
3 |
import re
|
4 |
import time
|
5 |
import logging
|
6 |
import os
|
|
|
|
|
7 |
from transformers import AutoTokenizer, GenerationConfig
|
8 |
from peft import AutoPeftModelForCausalLM
|
9 |
+
import gc
|
|
|
10 |
|
11 |
# -------- CONFIGURAÇÕES DE OTIMIZAÇÃO --------
|
12 |
os.environ["TOKENIZERS_PARALLELISM"] = "false"
|
|
|
15 |
torch.set_num_threads(2)
|
16 |
torch.set_num_interop_threads(1)
|
17 |
|
18 |
+
# -------- LOGGING CONFIG --------
|
19 |
logging.basicConfig(level=logging.INFO, format="%(asctime)s [%(levelname)s] %(message)s")
|
20 |
log = logging.getLogger("news-filter")
|
21 |
|
22 |
+
# -------- LOAD MODEL --------
|
23 |
model_name = "habulaj/filterinstruct180"
|
24 |
log.info("🚀 Carregando modelo e tokenizer...")
|
25 |
|
26 |
+
tokenizer = AutoTokenizer.from_pretrained(
|
27 |
+
model_name,
|
28 |
+
use_fast=True,
|
29 |
+
padding_side="left"
|
30 |
+
)
|
31 |
|
32 |
if tokenizer.pad_token is None:
|
33 |
tokenizer.pad_token = tokenizer.eos_token
|
|
|
37 |
device_map="cpu",
|
38 |
torch_dtype=torch.bfloat16,
|
39 |
low_cpu_mem_usage=True,
|
40 |
+
use_cache=True,
|
41 |
trust_remote_code=True
|
42 |
)
|
43 |
+
|
44 |
model.eval()
|
45 |
+
log.info("✅ Modelo carregado (eval mode).")
|
46 |
|
47 |
generation_config = GenerationConfig(
|
48 |
max_new_tokens=128,
|
|
|
52 |
use_cache=True,
|
53 |
eos_token_id=tokenizer.eos_token_id,
|
54 |
pad_token_id=tokenizer.eos_token_id,
|
55 |
+
no_repeat_ngram_size=2,
|
56 |
repetition_penalty=1.1,
|
57 |
length_penalty=1.0
|
58 |
)
|
59 |
|
60 |
+
# -------- FASTAPI INIT --------
|
61 |
app = FastAPI(title="News Filter JSON API")
|
62 |
|
63 |
@app.get("/")
|
64 |
def read_root():
|
65 |
return {"message": "News Filter JSON API is running!", "docs": "/docs"}
|
66 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
67 |
# -------- INFERÊNCIA --------
|
68 |
def infer_filter(title, content):
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
69 |
log.info(f"🧠 Inferência iniciada para: {title}")
|
70 |
start_time = time.time()
|
71 |
|
72 |
+
chat_prompt = build_chat_prompt(title, content)
|
73 |
+
|
74 |
+
inputs = tokenizer(
|
75 |
+
chat_prompt,
|
76 |
return_tensors="pt",
|
77 |
+
truncation=True,
|
78 |
+
max_length=512,
|
79 |
+
padding=False,
|
80 |
+
add_special_tokens=False
|
81 |
+
)
|
82 |
+
|
83 |
+
input_ids = inputs.input_ids
|
84 |
+
attention_mask = inputs.attention_mask
|
85 |
|
86 |
with torch.no_grad(), torch.inference_mode():
|
87 |
outputs = model.generate(
|
88 |
+
input_ids=input_ids,
|
89 |
+
attention_mask=attention_mask,
|
90 |
generation_config=generation_config,
|
91 |
+
num_return_sequences=1,
|
92 |
+
output_scores=False,
|
93 |
+
return_dict_in_generate=False
|
94 |
)
|
95 |
|
96 |
+
generated_tokens = outputs[0][len(input_ids[0]):]
|
97 |
+
generated = tokenizer.decode(
|
98 |
+
generated_tokens,
|
99 |
+
skip_special_tokens=True,
|
100 |
+
clean_up_tokenization_spaces=True
|
101 |
+
)
|
102 |
+
|
103 |
+
log.info("📤 Resultado gerado:")
|
104 |
+
log.info(generated)
|
105 |
+
|
106 |
+
json_result = extract_json(generated)
|
107 |
|
|
|
108 |
duration = time.time() - start_time
|
109 |
log.info(f"✅ JSON extraído em {duration:.2f}s")
|
110 |
+
|
111 |
+
# Limpeza de memória
|
112 |
+
del outputs, generated_tokens, inputs
|
113 |
+
gc.collect()
|
114 |
+
|
115 |
+
if json_result:
|
116 |
+
return json_result
|
117 |
+
else:
|
118 |
+
raise HTTPException(status_code=404, detail="Unable to extract JSON from model output.")
|
119 |
+
|
120 |
+
def build_chat_prompt(title: str, content: str) -> str:
|
121 |
+
return f"""<|begin_of_text|><|start_header_id|>user<|end_header_id|>
|
122 |
+
Analyze the news title and content, and return the filters in JSON format with the defined fields.
|
123 |
+
|
124 |
+
Please respond ONLY with the JSON filter, do NOT add any explanations, system messages, or extra text.
|
125 |
+
|
126 |
+
Title: "{title}"
|
127 |
+
Content: "{content}"<|eot_id|><|start_header_id|>assistant<|end_header_id|>"""
|
128 |
|
129 |
def extract_json(text):
|
130 |
+
match = re.search(r'\{[^{}]*(?:\{[^{}]*\}[^{}]*)*\}', text, re.DOTALL)
|
131 |
if match:
|
132 |
+
json_text = match.group(0)
|
133 |
+
|
134 |
+
# Conversões comuns
|
135 |
+
json_text = re.sub(r"'", '"', json_text)
|
136 |
+
json_text = re.sub(r'\bTrue\b', 'true', json_text)
|
137 |
+
json_text = re.sub(r'\bFalse\b', 'false', json_text)
|
138 |
+
json_text = re.sub(r",\s*}", "}", json_text)
|
139 |
+
json_text = re.sub(r",\s*]", "]", json_text)
|
140 |
+
return json_text.strip()
|
141 |
+
return text
|
142 |
+
|
143 |
+
# -------- API ROUTE --------
|
144 |
+
@app.get("/filter")
|
145 |
+
def get_filter(
|
146 |
+
title: str = Query(..., description="News title"),
|
147 |
+
content: str = Query(..., description="News content")
|
148 |
+
):
|
149 |
+
try:
|
150 |
+
json_output = infer_filter(title, content)
|
151 |
+
import json
|
152 |
+
try:
|
153 |
+
parsed = json.loads(json_output)
|
154 |
+
return {"result": parsed}
|
155 |
+
except json.JSONDecodeError as e:
|
156 |
+
log.error(f"❌ Erro ao parsear JSON: {e}")
|
157 |
+
return {"result": json_output, "warning": "JSON returned as string due to parsing error"}
|
158 |
+
except HTTPException as e:
|
159 |
+
raise e
|
160 |
+
except Exception as e:
|
161 |
+
log.exception("❌ Erro inesperado:")
|
162 |
+
raise HTTPException(status_code=500, detail="Internal server error during inference.")
|
163 |
+
|
164 |
+
@app.on_event("startup")
|
165 |
+
async def warmup():
|
166 |
+
log.info("🔥 Executando warmup...")
|
167 |
+
try:
|
168 |
+
infer_filter("Test title", "Test content")
|
169 |
+
log.info("✅ Warmup concluído.")
|
170 |
+
except Exception as e:
|
171 |
+
log.warning(f"⚠️ Warmup falhou: {e}")
|