[Refactor] Separate sequence and token pooling types (#32026)

Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
2026-01-10 12:53:24 +08:00
parent 52d428295d
commit 583a90e005
42 changed files with 324 additions and 204 deletions
--- a/tests/models/language/pooling_mteb_test/test_qwen3_reranker.py
+++ b/tests/models/language/pooling_mteb_test/test_qwen3_reranker.py
@@ -27,7 +27,7 @@ RERANK_MODELS = [
        architecture="Qwen3ForSequenceClassification",
        hf_overrides=qwen3_reranker_hf_overrides,
        chat_template_name="qwen3_reranker.jinja",
-        pooling_type="LAST",
+        seq_pooling_type="LAST",
        attn_type="decoder",
        is_prefix_caching_supported=True,
        is_chunked_prefill_supported=True,