diff options
| author | Douglas Hanley <thesecretaryofwar@gmail.com> | 2024-03-03 04:40:27 -0600 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2024-03-03 12:40:27 +0200 |
| commit | 475df1d6cf817060028d3ff763cb8097d4ec40d6 (patch) | |
| tree | 5cad43f149f24b7b3f40604b78b7971e458aa309 /common/common.h | |
| parent | 87c2e8b2797860a06af3d6c06b8488a8ff1a09ab (diff) | |
llama : allow for user specified embedding pooling type (#5849)
* allow for user specified pooling type
* llama : use enum types over int
---------
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
Diffstat (limited to 'common/common.h')
| -rw-r--r-- | common/common.h | 7 |
1 files changed, 5 insertions, 2 deletions
diff --git a/common/common.h b/common/common.h index ab62bdb8..d3682b7a 100644 --- a/common/common.h +++ b/common/common.h @@ -76,8 +76,11 @@ struct gpt_params { float yarn_beta_slow = 1.0f; // YaRN high correction dim int32_t yarn_orig_ctx = 0; // YaRN original context length float defrag_thold = -1.0f; // KV cache defragmentation threshold - int32_t rope_scaling_type = LLAMA_ROPE_SCALING_TYPE_UNSPECIFIED; - ggml_numa_strategy numa = GGML_NUMA_STRATEGY_DISABLED; + + ggml_numa_strategy numa = GGML_NUMA_STRATEGY_DISABLED; + + llama_rope_scaling_type rope_scaling_type = LLAMA_ROPE_SCALING_TYPE_UNSPECIFIED; + llama_pooling_type pooling_type = LLAMA_POOLING_TYPE_UNSPECIFIED; // pooling type for embeddings // // sampling parameters struct llama_sampling_params sparams; |