Spaces:
Running
on
Zero
Running
on
Zero
cutechicken
commited on
Commit
•
58e272a
1
Parent(s):
85ff42c
Update app.py
Browse files
app.py
CHANGED
@@ -6,62 +6,79 @@ import os
|
|
6 |
from threading import Thread
|
7 |
import random
|
8 |
from datasets import load_dataset
|
|
|
|
|
|
|
|
|
|
|
|
|
9 |
|
10 |
HF_TOKEN = os.environ.get("HF_TOKEN", None)
|
11 |
MODEL_ID = "CohereForAI/c4ai-command-r7b-12-2024"
|
12 |
MODELS = os.environ.get("MODELS")
|
13 |
MODEL_NAME = MODEL_ID.split("/")[-1]
|
14 |
|
15 |
-
|
16 |
-
|
17 |
-
CSS = """
|
18 |
-
.duplicate-button {
|
19 |
-
margin: auto !important;
|
20 |
-
color: white !important;
|
21 |
-
background: black !important;
|
22 |
-
border-radius: 100vh !important;
|
23 |
-
}
|
24 |
-
h3 {
|
25 |
-
text-align: center;
|
26 |
-
}
|
27 |
-
.chatbox .messages .message.user {
|
28 |
-
background-color: #e1f5fe;
|
29 |
-
}
|
30 |
-
.chatbox .messages .message.bot {
|
31 |
-
background-color: #eeeeee;
|
32 |
-
}
|
33 |
-
"""
|
34 |
|
35 |
-
#
|
36 |
-
|
37 |
-
|
38 |
-
torch_dtype=torch.bfloat16,
|
39 |
-
device_map="auto",
|
40 |
-
)
|
41 |
-
tokenizer = AutoTokenizer.from_pretrained(MODEL_ID)
|
42 |
|
43 |
-
#
|
44 |
-
|
45 |
-
|
46 |
|
47 |
-
|
48 |
-
|
49 |
-
|
50 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
51 |
|
52 |
@spaces.GPU
|
53 |
def stream_chat(message: str, history: list, temperature: float, max_new_tokens: int, top_p: float, top_k: int, penalty: float):
|
54 |
print(f'message is - {message}')
|
55 |
print(f'history is - {history}')
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
56 |
conversation = []
|
57 |
for prompt, answer in history:
|
58 |
-
conversation.extend([
|
59 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
60 |
|
61 |
input_ids = tokenizer.apply_chat_template(conversation, tokenize=False, add_generation_prompt=True)
|
62 |
inputs = tokenizer(input_ids, return_tensors="pt").to(0)
|
63 |
|
64 |
-
streamer = TextIteratorStreamer(tokenizer, timeout
|
65 |
|
66 |
generate_kwargs = dict(
|
67 |
inputs,
|
|
|
6 |
from threading import Thread
|
7 |
import random
|
8 |
from datasets import load_dataset
|
9 |
+
from sentence_transformers import SentenceTransformer
|
10 |
+
from sklearn.metrics.pairwise import cosine_similarity
|
11 |
+
import numpy as np
|
12 |
+
|
13 |
+
# GPU 메모리 관리
|
14 |
+
torch.cuda.empty_cache()
|
15 |
|
16 |
HF_TOKEN = os.environ.get("HF_TOKEN", None)
|
17 |
MODEL_ID = "CohereForAI/c4ai-command-r7b-12-2024"
|
18 |
MODELS = os.environ.get("MODELS")
|
19 |
MODEL_NAME = MODEL_ID.split("/")[-1]
|
20 |
|
21 |
+
# 임베딩 모델 로드
|
22 |
+
embedding_model = SentenceTransformer('sentence-transformers/xlm-r-100langs-bert-base-nli-stsb-mean-tokens')
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
23 |
|
24 |
+
# 위키피디아 데이터셋 로드
|
25 |
+
wiki_dataset = load_dataset("lcw99/wikipedia-korean-20240501-1million-qna")
|
26 |
+
print("Wikipedia dataset loaded:", wiki_dataset)
|
|
|
|
|
|
|
|
|
27 |
|
28 |
+
# 데이터셋의 질문들을 임베딩
|
29 |
+
questions = wiki_dataset['train']['question'][:10000] # 처음 10000개만 사용
|
30 |
+
question_embeddings = embedding_model.encode(questions, convert_to_tensor=True)
|
31 |
|
32 |
+
def find_relevant_context(query, top_k=3):
|
33 |
+
# 쿼리 임베딩
|
34 |
+
query_embedding = embedding_model.encode(query, convert_to_tensor=True)
|
35 |
+
|
36 |
+
# 코사인 유사도 계산
|
37 |
+
similarities = cosine_similarity(
|
38 |
+
query_embedding.cpu().numpy().reshape(1, -1),
|
39 |
+
question_embeddings.cpu().numpy()
|
40 |
+
)[0]
|
41 |
+
|
42 |
+
# 가장 유사한 질문들의 인덱스
|
43 |
+
top_indices = np.argsort(similarities)[-top_k:][::-1]
|
44 |
+
|
45 |
+
# 관련 컨텍스트 추출
|
46 |
+
relevant_contexts = []
|
47 |
+
for idx in top_indices:
|
48 |
+
relevant_contexts.append({
|
49 |
+
'question': questions[idx],
|
50 |
+
'answer': wiki_dataset['train']['answer'][idx]
|
51 |
+
})
|
52 |
+
|
53 |
+
return relevant_contexts
|
54 |
|
55 |
@spaces.GPU
|
56 |
def stream_chat(message: str, history: list, temperature: float, max_new_tokens: int, top_p: float, top_k: int, penalty: float):
|
57 |
print(f'message is - {message}')
|
58 |
print(f'history is - {history}')
|
59 |
+
|
60 |
+
# RAG: 관련 컨텍스트 찾기
|
61 |
+
relevant_contexts = find_relevant_context(message)
|
62 |
+
context_prompt = "\n\n관련 참고 정보:\n"
|
63 |
+
for ctx in relevant_contexts:
|
64 |
+
context_prompt += f"Q: {ctx['question']}\nA: {ctx['answer']}\n\n"
|
65 |
+
|
66 |
+
# 대화 히스토리 구성
|
67 |
conversation = []
|
68 |
for prompt, answer in history:
|
69 |
+
conversation.extend([
|
70 |
+
{"role": "user", "content": prompt},
|
71 |
+
{"role": "assistant", "content": answer}
|
72 |
+
])
|
73 |
+
|
74 |
+
# 컨텍스트를 포함한 최종 프롬프트 구성
|
75 |
+
final_message = context_prompt + "\n현재 질문: " + message
|
76 |
+
conversation.append({"role": "user", "content": final_message})
|
77 |
|
78 |
input_ids = tokenizer.apply_chat_template(conversation, tokenize=False, add_generation_prompt=True)
|
79 |
inputs = tokenizer(input_ids, return_tensors="pt").to(0)
|
80 |
|
81 |
+
streamer = TextIteratorStreamer(tokenizer, timeout
|
82 |
|
83 |
generate_kwargs = dict(
|
84 |
inputs,
|