ggerganov/llama.cpp

Pull Requests Commits

gguf-py : avoid adding duplicate tensor mappings for Jamba

compilade committed 164 days ago

4d6a179c

memory : avoid referring to KV in recurrent cache logs

compilade committed 164 days ago

452207f3

model : make falcon-h1 use shared mamba2 layer builder

compilade committed 164 days ago

7f3955a0

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

compilade committed 164 days ago

a60a24be

ggml : prevent integer overflow in gguf tensor size calculation (#14595)

Yuuoniy committed 164 days ago

Verified 26a48ad6

model : add skt/A.X-4.0 model vocabulary (#14589)

Bing-su committed 164 days ago

Verified ffd59e7d

llama : remove unintended whitespace (#14592)

CISC committed 164 days ago

Verified 10555459

model : add support for Falcon-H1 family (#14534)

ibrahimkhadraoui committed 164 days ago

Verified 04655063

convert : fix smollm3 jinja template (#14586)

ngxson committed 164 days ago

Verified 20b7bf8a

model : use ggml_swiglu_split for Mamba

compilade committed 165 days ago

f7c7a926

model : remove unnecessary prefix for tensor loading constants

compilade committed 165 days ago

2f39cd7b

jamba : remove redundant nullptr initializations

compilade committed 165 days ago

db5ff0cc

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

compilade committed 165 days ago

b0b280ea

vulkan: optimize flash attention split_k_reduce (#14554)

jeffbolznv committed 165 days ago

Verified 6efcd659

model : fix hunyuan moe chat template (#14584)

stevenkuang-tencent committed 165 days ago

Verified 699f4392

model : add SmolLM3 (#14581)

ngxson committed 165 days ago

Verified 08382869

memory : fix broken batch splits for recurrent cache (#14575)

compilade committed 165 days ago

Verified bb4f7a9e

vulkan : fix rope with partial rotation and non-cont src (#14582)

jeffbolznv committed 165 days ago

Verified b8eeb874

server: Add ability to mount server at prefix (#14544)

oluwandabira committed 165 days ago

Verified 17a1f0d2

model : add hunyuan moe (#14425)

ngxson committed 165 days ago

Verified 8f22dc0a

vulkan: increase timeout for CI (#14574)

jeffbolznv committed 165 days ago

Verified 53903ae6

cuda : fix rope with partial rotation and non-cont src (#14580)

ggerganov committed 165 days ago

Verified 4d0dcd4a

CUDA: add bilinear interpolation for upscale (#14563)

am17an committed 165 days ago

Verified 75c91de6

musa: fix build warnings (unused variable) (#14561)

yeahdongcn committed 165 days ago

Verified 68155c66

llama : fix incorrect minicpm3 v_states shape (#14571)

CISC committed 165 days ago

Verified e1a70590

llama : remove ggml_cont where possible (#14568)

CISC committed 166 days ago

Verified 12f55c30

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

compilade committed 166 days ago

f7163582

CUDA: add bf16 and i32 to getrows (#14529)

am17an committed 166 days ago

Verified b9c3eefd

vulkan: increase LOAD_VEC_A to 8 (IQ1/IQ2) or 4 (IQ3) (#14485)

netrunnereve committed 167 days ago

Verified 6491d6e4

vulkan: fix rms_norm+mul fusion (#14545)

jeffbolznv committed 167 days ago

Verified e592be15

Older