Update src/streamlit_app.py
Browse files- src/streamlit_app.py +6 -5
src/streamlit_app.py
CHANGED
@@ -73,11 +73,7 @@ def extract_numeric_ram(ram) -> Optional[int]:
|
|
73 |
|
74 |
# Streamlined LLM database with popular models and download sizes
|
75 |
LLM_DATABASE = {
|
76 |
-
|
77 |
-
"fp16": {"multiplier": 1.0, "label": "FP16 (Full)", "description": "Best quality, largest size"},
|
78 |
-
"8bit": {"multiplier": 0.5, "label": "8-bit", "description": "Good quality, 50% smaller"},
|
79 |
-
"4bit": {"multiplier": 0.25, "label": "4-bit", "description": "Decent quality, 75% smaller"}
|
80 |
-
}
|
81 |
"ultra_low": { # ≤2GB
|
82 |
"general": [
|
83 |
{"name": "TinyLlama-1.1B-Chat", "size": "637MB", "description": "Compact chat model"},
|
@@ -187,6 +183,11 @@ LLM_DATABASE = {
|
|
187 |
]
|
188 |
}
|
189 |
}
|
|
|
|
|
|
|
|
|
|
|
190 |
|
191 |
# Enhanced LLM recommendation with performance tiers
|
192 |
def recommend_llm(ram_str) -> Tuple[str, str, str, Dict[str, List[Dict]]]:
|
|
|
73 |
|
74 |
# Streamlined LLM database with popular models and download sizes
|
75 |
LLM_DATABASE = {
|
76 |
+
|
|
|
|
|
|
|
|
|
77 |
"ultra_low": { # ≤2GB
|
78 |
"general": [
|
79 |
{"name": "TinyLlama-1.1B-Chat", "size": "637MB", "description": "Compact chat model"},
|
|
|
183 |
]
|
184 |
}
|
185 |
}
|
186 |
+
QUANTIZATION_INFO = {
|
187 |
+
"fp16": {"multiplier": 1.0, "label": "FP16 (Full)", "description": "Best quality, largest size"},
|
188 |
+
"8bit": {"multiplier": 0.5, "label": "8-bit", "description": "Good quality, 50% smaller"},
|
189 |
+
"4bit": {"multiplier": 0.25, "label": "4-bit", "description": "Decent quality, 75% smaller"}
|
190 |
+
}
|
191 |
|
192 |
# Enhanced LLM recommendation with performance tiers
|
193 |
def recommend_llm(ram_str) -> Tuple[str, str, str, Dict[str, List[Dict]]]:
|