model-conversion : use CONVERTED_EMBEDDING_MODEL for embedding_verify_logits (#18079)

This commit updates the embedding model verification script to use the
CONVERTED_EMBEDDING_MODEL environment variable instead of using the
EMBEDDING_MODEL_PATH (the original embedding model path) as the basis
for the converted model file name.

The motivation for this that currently if the converted embedding model
file name differs from the original embedding model directory/name the
verification script will look for the wrong .bin files that were
generating when running the models.

Author: danbev

examples/model-conversion/scripts/embedding/compare-embeddings-logits.sh

@@ -34,8 +34,11 @@ done
 MODEL_PATH="${MODEL_PATH:-"$EMBEDDING_MODEL_PATH"}"
 MODEL_NAME="${MODEL_NAME:-$(basename "$MODEL_PATH")}"
 
+CONVERTED_MODEL_PATH="${CONVERTED_EMBEDDING_PATH:-"$CONVERTED_EMBEDDING_MODEL"}"
+CONVERTED_MODEL_NAME="${CONVERTED_MODEL_NAME:-$(basename "$CONVERTED_MODEL_PATH" .gguf)}"
+
 if [ -t 0 ]; then
-    CPP_EMBEDDINGS="data/llamacpp-${MODEL_NAME}-embeddings.bin"
+    CPP_EMBEDDINGS="data/llamacpp-${CONVERTED_MODEL_NAME}-embeddings.bin"
 else
     # Process piped JSON data and convert to binary (matching logits.cpp format)
     TEMP_FILE=$(mktemp /tmp/tmp.XXXXXX.binn)