vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

llama : grouped-query attention + LLaMAv2 70B support (#2276)

Browse source 
* CUDA: GQA implementation

* llama : support for GQA and LLaMAv2 70B

ggml-ci

* py : fix hparams parsing (if-else blocks)

ggml-ci

* py : oh boy ..

ggml-ci

* help : fix gqa value for 70B

ggml-ci

---------

Co-authored-by: JohannesGaessler <johannesg@5d6.de>
This commit is contained in:
Georgi Gerganov 2023-07-23 15:09:47 +03:00 • committed by GitHub
parent 1d0824b247
commit e76d630df1
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
7 changed files with 215 additions and 108 deletions
Download patch file Download diff file Expand all files Collapse all files

11
llama.h
View file

@ -83,11 +83,12 @@ extern "C" {
typedef void (*llama_progress_callback)(float progress, void *ctx);
struct llama_context_params {
uint32_t seed; // RNG seed, -1 for random
int32_t n_ctx; // text context
int32_t n_batch; // prompt processing batch size
int32_t n_gpu_layers; // number of layers to store in VRAM
int32_t main_gpu; // the GPU that is used for scratch and small tensors
uint32_t seed; // RNG seed, -1 for random
int32_t n_ctx; // text context
int32_t n_batch; // prompt processing batch size
int32_t n_gqa; // grouped-query attention (TEMP - will be moved to model hparams)
int32_t n_gpu_layers; // number of layers to store in VRAM
int32_t main_gpu; // the GPU that is used for scratch and small tensors
const float * tensor_split; // how to split layers across multiple GPUs (size: LLAMA_MAX_DEVICES)