ggerganov/llama.cpp

Pull Requests Commits

Merge branch 'master' into gg/ggml_scale

ggerganov committed 1 year ago

Verified ab1b7516

llama : initial ggml-backend integration (#4520)

slaren committed 1 year ago

Verified d232aca5

llama : allow getting n_batch from llama_context in c api (#4540)

MarcusDunn committed 1 year ago

Verified 31f27758

metal : fix `ggml_metal_log` vargs (#4373)

finnvoor committed 1 year ago

Verified 56fa5081

cuda : ROCm AMD Unified Memory Architecture (UMA) handling (#4449)

ekg committed 1 year ago

Verified 0f630fbc

tests : fix test-grad0

ggerganov committed 1 year ago

Verified b784f881

ggml-cuda: Fix HIP build by adding define for __trap (#4569)

arlo-phoenix committed 1 year ago

Verified 562cf222

ggml : fix CPU implementation

ggerganov committed 1 year ago

Verified 36c3f41f

ggml : change ggml_scale to take a float instead of tensor

ggerganov committed 1 year ago

Verified 199f6bdc

common : remove incorrect --model-draft default (#4568)

cebtenzzre committed 1 year ago

Verified 8fe03ffd

CUDA: mul_mat_id always on GPU for batches >= 32 (#4553)

JohannesGaessler committed 1 year ago

Verified 91544948

readme : update coding guidelines

ggerganov committed 1 year ago

Verified c083718c

py : open merges file as 'utf-8' (#4566)

howlger committed 1 year ago

Verified 880e3522

cuda : better error message for ggml_get_rows (#4561)

bobqianic committed 1 year ago

Verified 66f35a2f

cuda : replace asserts in wrong architecture checks with __trap (#4556)

slaren committed 1 year ago

Verified 13988239

llama : disable per-tensor info prints on model load (#4562)

JohannesGaessler committed 1 year ago

Verified d3223afd

Fix access violation in ggml_cuda_free_data if tensor->extra is NULL (#4554)

LoganDark committed 1 year ago

Verified 1d7a1912

CUDA: Faster Mixtral prompt processing (#4538)

JohannesGaessler committed 1 year ago

Verified 799fc226

ggml : fixed check for _MSC_VER (#4535)

es0m committed 1 year ago

Verified 328b83de

ggml-cuda: Fix HIP build (#4528)

arlo-phoenix committed 1 year ago

Verified a7aee47b

llama.swiftui : add tinyllama 1.1B F16

ggerganov committed 1 year ago

Verified 0e18b2e7

llama.swiftui : add more models

ggerganov committed 1 year ago

Verified 6ff39b12

llama : add phi-2 + fix NeoX rope + ggml_mul_mat_set_prec (#4490)

ebeyabraham committed 1 year ago

Verified b9e74f9b

llama : fix try_override for bool_value which always return true (#4519)

hankcs committed 1 year ago

Verified 3c04bf6d

decode : fix logits_valid for legacy API (#4516)

cebtenzzre committed 1 year ago

Verified 2994f0c5

readme : update hot topics

ggerganov committed 1 year ago

Verified b1306c43

llama.swiftui : add bench functionality (#4483)

ggerganov committed 1 year ago

Verified 800a489e

gguf-py : fail fast on nonsensical special token IDs (#4489)

cebtenzzre committed 1 year ago

Verified f7f468a9

build : Check the ROCm installation location (#4485)

Su3h7aM committed 1 year ago

Verified 919c4066

finetune : keep allocs alive until all allocations are done (#4486)

slaren committed 1 year ago

Verified 45668633

Older