From 06e85401b0f265f8ffa76144d31c1c7fdca45b6b Mon Sep 17 00:00:00 2001 From: Georgi Gerganov Date: Tue, 17 Dec 2024 10:56:20 +0200 Subject: [PATCH] server : output embeddings for all tokens when pooling = none ggml-ci --- examples/server/tests/utils.py | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/examples/server/tests/utils.py b/examples/server/tests/utils.py index 277125e88..da95c830b 100644 --- a/examples/server/tests/utils.py +++ b/examples/server/tests/utils.py @@ -275,7 +275,7 @@ class ServerPreset: return server @staticmethod - def bert_bge_small() -> ServerProcess: + def bert_bge_small(pooling = 'last') -> ServerProcess: server = ServerProcess() server.model_hf_repo = "ggml-org/models" server.model_hf_file = "bert-bge-small/ggml-model-f16.gguf" @@ -286,6 +286,7 @@ class ServerPreset: server.n_slots = 2 server.seed = 42 server.server_embeddings = True + server.pooling = pooling return server @staticmethod