ggerganov/llama.cpp

Pull Requests Commits

llama : use unused n_embd_k_gqa in k_shift

compilade committed 261 days ago

375de5b1

llama : fix mixed signedness comparison

compilade committed 261 days ago

5f62db79

convert_hf : fix Jamba conversion

compilade committed 261 days ago

9d3f44da

Merge branch 'master' into compilade/refactor-kv-cache

compilade committed 261 days ago

a03e32a3

llama : session saving and reloading for hybrid models

compilade committed 261 days ago

fcb889cf

llama : support RWKV v6 models (#8980)

MollySophia committed 261 days ago

Verified 8f1d81a0

Merge branch 'master' into compilade/refactor-kv-cache

compilade committed 262 days ago

bc320ef6

nix: fix CUDA build - replace deprecated autoAddOpenGLRunpathHook

enolan committed 262 days ago

a47667cf

sgemm : improved Q4_0 and Q8_0 performance via 4xN and Mx4 gemm (#8908)

Srihari-mcw committed 262 days ago

Verified ea5d7478

llama : fix typo in xcda_array_view comment [no ci] (#9132)

danbev committed 262 days ago

Verified 49271efb

llama : fix llama_split_mode enum values in main_gpu document (#9057)

kou committed 263 days ago

Verified 0ab30f8d

Correct typo run_llama2.sh > run-llama2.sh (#9149)

shou692199 committed 263 days ago

Verified cddae488

llava : the function "clip" should be int (#9237)

tc-mb committed 264 days ago

Verified 7ea8d80d

Threadpool: take 2 (#8672)

fmz committed 264 days ago

Verified 42c76d13

server : fix crash when error handler dumps invalid utf-8 json (#9195)

kaetemi committed 264 days ago

9f7d4bcf

flake.lock: Update (#9162)

ggerganov committed 265 days ago

Verified 1d1ccce6

docker : build images only once (#9225)

slaren committed 265 days ago

Verified 9fe94cca

docker : update CUDA images (#9213)

slaren committed 265 days ago

Verified 66b039a5

vulkan : fix build (#0)

ggerganov committed 266 days ago

20f1789d

ggerganov committed 266 days ago

231cff5f

Fix minicpm example directory (#9111)

xyb committed 266 days ago

Verified 3246fe84

llama : fix qs.n_attention_wv for DeepSeek-V2 (#9156)

compilade committed 266 days ago

Verified 78eb487b

server : add some missing env variables (#9116)

ngxson committed 266 days ago

Verified a77feb5d

llama : fix ChatGLM4 wrong shape (#9194)

CausalLM committed 266 days ago

Verified 2e59d61c

llama : fix llama3.1 rope_freqs not respecting custom head_dim (#9141)

nyxkrage committed 266 days ago

Verified 75e1dbba

common : Update stb_image.h to latest version (#9161)

arch-btw committed 266 days ago

Verified ad76569f

ggml : do not crash when quantizing q4_x_x with an imatrix (#9192)

slaren committed 267 days ago

Verified 7d787ed9

metal : separate scale and mask from QKT in FA kernel (#9189)

ggerganov committed 267 days ago

Verified 06658ad7

ggml : add SSM Metal kernels (#8546)

ggerganov committed 267 days ago

Verified fc18425b

tests : fix compile warnings for unreachable code (#9185)

ggerganov committed 267 days ago

Verified 879275ac

Older