omaryasserhassan commited on
Commit
778f3aa
·
verified ·
1 Parent(s): 8b2ac0c

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +4 -10
app.py CHANGED
@@ -3,7 +3,7 @@ import traceback
3
  from fastapi import FastAPI, HTTPException
4
  from pydantic import BaseModel
5
  from huggingface_hub import hf_hub_download
6
- from ctransformers import AutoModelForCausalLM
7
 
8
  # --- Config ---
9
  REPO_ID = "bartowski/Llama-3.2-3B-Instruct-GGUF"
@@ -24,25 +24,19 @@ def get_model():
24
  if _model is not None:
25
  return _model
26
 
27
- print("📥 Downloading model...")
28
  local_file = hf_hub_download(
29
  repo_id=REPO_ID,
30
  filename=FILENAME,
31
  cache_dir=CACHE_DIR,
32
  local_dir_use_symlinks=False,
33
  )
34
- local_dir = os.path.dirname(local_file)
35
- print("✅ Model downloaded at", local_file)
36
 
37
- print("🔄 Loading model into memory...")
38
- _model = AutoModelForCausalLM.from_pretrained(
39
- local_dir,
40
- model_file=FILENAME,
41
  model_type=MODEL_TYPE,
42
- gpu_layers=0, # CPU only
43
  threads=os.cpu_count() or 2
44
  )
45
- print("✅ Model loaded")
46
  return _model
47
 
48
  # --- Request Schema ---
 
3
  from fastapi import FastAPI, HTTPException
4
  from pydantic import BaseModel
5
  from huggingface_hub import hf_hub_download
6
+ from ctransformers import LLM
7
 
8
  # --- Config ---
9
  REPO_ID = "bartowski/Llama-3.2-3B-Instruct-GGUF"
 
24
  if _model is not None:
25
  return _model
26
 
 
27
  local_file = hf_hub_download(
28
  repo_id=REPO_ID,
29
  filename=FILENAME,
30
  cache_dir=CACHE_DIR,
31
  local_dir_use_symlinks=False,
32
  )
 
 
33
 
34
+ _model = LLM(
35
+ model=local_file, # direct file path
 
 
36
  model_type=MODEL_TYPE,
37
+ gpu_layers=0,
38
  threads=os.cpu_count() or 2
39
  )
 
40
  return _model
41
 
42
  # --- Request Schema ---