Update app.py
Browse files
app.py
CHANGED
|
@@ -8,7 +8,7 @@ MODEL_REPO = "Kezovic/iris-q4gguf-nitrallora"
|
|
| 8 |
MODEL_FILE = "Llama-3.2-1B-Instruct.Q4_K_M.gguf"
|
| 9 |
CONTEXT_WINDOW = 2048
|
| 10 |
MAX_NEW_TOKENS = 400
|
| 11 |
-
TEMPERATURE =
|
| 12 |
|
| 13 |
# --- Model Loading ---
|
| 14 |
def load_llm():
|
|
@@ -106,7 +106,7 @@ def analyze_poem(poem_text, progress=gr.Progress()):
|
|
| 106 |
output_poet = llm_analyst(
|
| 107 |
prompt=poet_prompt,
|
| 108 |
max_tokens=200, # Keep this response short
|
| 109 |
-
temperature=
|
| 110 |
# Minimal stop sequences for clean output
|
| 111 |
echo=False
|
| 112 |
)
|
|
|
|
| 8 |
MODEL_FILE = "Llama-3.2-1B-Instruct.Q4_K_M.gguf"
|
| 9 |
CONTEXT_WINDOW = 2048
|
| 10 |
MAX_NEW_TOKENS = 400
|
| 11 |
+
TEMPERATURE = 1.5 # Note: min_p = 0.1 is kept as requested in the configuration from the previous prompt
|
| 12 |
|
| 13 |
# --- Model Loading ---
|
| 14 |
def load_llm():
|
|
|
|
| 106 |
output_poet = llm_analyst(
|
| 107 |
prompt=poet_prompt,
|
| 108 |
max_tokens=200, # Keep this response short
|
| 109 |
+
temperature=1.5,
|
| 110 |
# Minimal stop sequences for clean output
|
| 111 |
echo=False
|
| 112 |
)
|