bert : add tests + fix quantization (#5475)

* llama : do not quantize pos embd and token type tensors

* ci : add BERT tests

ggml-ci

* ci : do not do BERT tests on low-perf nodes

ggml-ci

1 files changed, 5 insertions, 1 deletions

diff --git a/llama.cpp b/llama.cpp
index d316d067..6dce392d 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -10444,7 +10444,11 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
         quantize &= !params->only_copy;
 
         // do not quantize expert gating tensors
-        quantize &= name.find("ffn_gate_inp.weight") == std::string::npos;
+        quantize &= name != LLM_TN(model.arch)(LLM_TENSOR_FFN_GATE_INP, "weight");
+
+        // do not quantize positional embeddings and token types (BERT)
+        quantize &= name != LLM_TN(model.arch)(LLM_TENSOR_POS_EMBD,    "weight");
+        quantize &= name != LLM_TN(model.arch)(LLM_TENSOR_TOKEN_TYPES, "weight");
 
         enum ggml_type new_type;
         void * new_data;