Update app.py
Browse files
app.py
CHANGED
|
@@ -33,13 +33,11 @@ warnings.filterwarnings("ignore")
|
|
| 33 |
print("Start")
|
| 34 |
import subprocess
|
| 35 |
|
| 36 |
-
import subprocess
|
| 37 |
-
|
| 38 |
try:
|
| 39 |
subprocess.run([
|
| 40 |
"huggingface-cli", "download",
|
| 41 |
-
"
|
| 42 |
-
"
|
| 43 |
"--local-dir", "./models",
|
| 44 |
"--local-dir-use-symlinks", "False"
|
| 45 |
], check=True)
|
|
@@ -65,7 +63,7 @@ embeddings = HuggingFaceEmbeddings(
|
|
| 65 |
# Load Mistral GGUF via llama.cpp (CPU optimized)
|
| 66 |
# ------------------------------
|
| 67 |
llm_cpp = Llama(
|
| 68 |
-
model_path="./models/Phi-3-mini-4k-instruct-gguf",
|
| 69 |
n_ctx=2048,
|
| 70 |
n_threads=4, # Adjust based on your CPU cores
|
| 71 |
n_gpu_layers=0, # Force CPU-only
|
|
|
|
| 33 |
print("Start")
|
| 34 |
import subprocess
|
| 35 |
|
|
|
|
|
|
|
| 36 |
try:
|
| 37 |
subprocess.run([
|
| 38 |
"huggingface-cli", "download",
|
| 39 |
+
"microsoft/Phi-3-mini-4k-instruct-gguf",
|
| 40 |
+
"Phi-3-mini-4k-instruct-q4.gguf",
|
| 41 |
"--local-dir", "./models",
|
| 42 |
"--local-dir-use-symlinks", "False"
|
| 43 |
], check=True)
|
|
|
|
| 63 |
# Load Mistral GGUF via llama.cpp (CPU optimized)
|
| 64 |
# ------------------------------
|
| 65 |
llm_cpp = Llama(
|
| 66 |
+
model_path="./models/Phi-3-mini-4k-instruct-q4.gguf",
|
| 67 |
n_ctx=2048,
|
| 68 |
n_threads=4, # Adjust based on your CPU cores
|
| 69 |
n_gpu_layers=0, # Force CPU-only
|