Spaces:
Configuration error
Configuration error
File size: 3,423 Bytes
78f5ea0 |
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 |
import torch
from transformers import AutoModelForCausalLM, AutoTokenizer
from autogen import AssistantAgent, UserProxyAgent
from config import LLM_MODEL, CONFIDENCE_THRESHOLD, VECTORSTORE_DIR
from rag import RAGAgent
import os
import sys
BASE_DIR = os.path.dirname(os.path.abspath(__file__))
if BASE_DIR not in sys.path:
sys.path.insert(0, BASE_DIR)
# Load BioMistral once
class BioMistralModel:
def __init__(self, model_name=LLM_MODEL, device=None):
print(f"[BioMistralModel] Loading model: {model_name}")
self.device = device or ("cuda" if torch.cuda.is_available() else "cpu")
self.tokenizer = AutoTokenizer.from_pretrained(model_name)
self.model = AutoModelForCausalLM.from_pretrained(
model_name,
torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32,
device_map="auto"
)
def generate_answer(self, query: str) -> str:
prompt = f"You are a helpful bioinformatics tutor. Answer clearly:\n\nQuestion: {query}\nAnswer:"
inputs = self.tokenizer(prompt, return_tensors="pt").to(self.device)
with torch.no_grad():
outputs = self.model.generate(
**inputs,
max_new_tokens=512,
do_sample=True,
top_p=0.95,
temperature=0.7,
pad_token_id=self.tokenizer.eos_token_id
)
text = self.tokenizer.decode(outputs[0], skip_special_tokens=True)
return text.split("Answer:", 1)[-1].strip()
# Formatting Agent
class FormattingAgent(AssistantAgent):
def __init__(self, name="FormattingAgent", **kwargs):
super().__init__(name=name, **kwargs)
def format_text(self, text: str) -> str:
cleaned = " ".join(text.split())
if cleaned:
cleaned = cleaned[0].upper() + cleaned[1:]
return cleaned
# Tutor Agent
class TutorAgent(AssistantAgent):
def __init__(self, name="TutorAgent", **kwargs):
super().__init__(name=name, **kwargs)
self.model = BioMistralModel()
self.format_agent = FormattingAgent()
self.rag_agent = RAGAgent(vectorstore_dir=str(VECTORSTORE_DIR)) # Exists but unused
def process_query(self, query: str) -> str:
print(f"[TutorAgent] Received query: {query}")
# Direct BioMistral answer
answer = self.model.generate_answer(query)
confidence = self.estimate_confidence(answer)
print(f"[TutorAgent] Confidence: {confidence:.2f}")
if confidence < CONFIDENCE_THRESHOLD:
print("[TutorAgent] Confidence low, but still using BioMistral (RAG unused).")
# Format with FormattingAgent
final_answer = self.format_agent.format_text(answer)
return final_answer
def estimate_confidence(self, answer: str) -> float:
"""
Dummy confidence estimator — could be replaced with actual logic.
"""
length = len(answer.strip())
if length > 100:
return 0.9
elif length > 50:
return 0.75
else:
return 0.5
#
# User Agent
class BioUser(UserProxyAgent):
def __init__(self, name="BioUser", **kwargs):
super().__init__(name=name, **kwargs)
|