Set temperature

Author: Lorenzo Toniazzi

tests/test_lora_conversion_and_inference.sh

@@ -39,16 +39,16 @@ run_conversion_and_inference_lora() {
     # Run inference
     echo "Running llama-cli without lora for $model_name with size $size_matrix..."
     OUTPUT_BASE=$(llama-cli -m $MODELS_REPO/$model_name/size=$size_matrix/base/Base-F32.gguf \
-    -p "Look again at that dot." -n 50 --seed 42)
+    -p "Look again at that dot." -n 50 --seed 42 --temp 0)
 
     echo "Running llama-cli with lora for $model_name with size $size_matrix..."
     OUTPUT_LORA_HOT=$(llama-cli -m $MODELS_REPO/$model_name/size=$size_matrix/base/Base-F32.gguf \
     --lora $MODELS_REPO/$model_name/size=$size_matrix/lora/Lora-F32-LoRA.gguf \
-    -p "I see a little silhouetto" -n 50 --seed 42)
+    -p "I see a little silhouetto" -n 50 --seed 42 --temp 0)
 
     echo "Running llama-cli with exported lora for $model_name with size $size_matrix..."
     OUTPUT_LORA_MERGED=$(llama-cli -m $MODELS_REPO/$model_name/size=$size_matrix/base/Base-F32-lora-merged.gguf \
-    -p "I see a little silhouetto" -n 50 --seed 42)
+    -p "I see a little silhouetto" -n 50 --seed 42 --temp 0)
 
     # Store the results in the regular array
     results+=("