sync from b7516

examples/model-conversion/scripts/embedding/run-converted-model.sh

@@ -5,7 +5,7 @@ set -e
 # Parse command line arguments
 CONVERTED_MODEL=""
 PROMPTS_FILE=""
-EMBD_NORMALIZE="2"
+USE_POOLING=""
 
 while [[ $# -gt 0 ]]; do
     case $1 in
@@ -13,9 +13,9 @@ while [[ $# -gt 0 ]]; do
             PROMPTS_FILE="$2"
             shift 2
             ;;
-        --embd-normalize)
-            EMBD_NORMALIZE="$2"
-            shift 2
+        --pooling)
+            USE_POOLING="1"
+            shift
             ;;
         *)
             if [ -z "$CONVERTED_MODEL" ]; then
@@ -50,5 +50,10 @@ fi
 
 echo $CONVERTED_MODEL
 
-cmake --build ../../build --target llama-debug -j8
-../../build/bin/llama-debug -m "$CONVERTED_MODEL" --embedding -p "$PROMPT" --save-logits --embd-normalize $EMBD_NORMALIZE
+cmake --build ../../build --target llama-logits -j8
+# TODO: update logits.cpp to accept a --file/-f option for the prompt
+if [ -n "$USE_POOLING" ]; then
+    ../../build/bin/llama-logits -m "$CONVERTED_MODEL" -embd-mode -pooling "$PROMPT"
+else
+    ../../build/bin/llama-logits -m "$CONVERTED_MODEL" -embd-mode "$PROMPT"
+fi