ggerganov/llama.cpp

Pull Requests Commits

llama-bench : skip repeated values in consecutive lines

slaren committed 203 days ago

4fc8673d

llama : adjust default context size + print warnings (#10136)

ggerganov committed 203 days ago

Verified 1926d6e3

simple-chat : only add bos on first prompt (#10129)

slaren committed 203 days ago

Verified b634f8a2

convert-lora : make `--base` optional (#10110)

ngxson committed 203 days ago

Verified 7554aa46

llama : add simple-chat example (#10124)

slaren committed 204 days ago

Verified a6744e43

llama : use smart pointers for ggml resources (#10117)

slaren committed 204 days ago

Verified e991e312

vulkan : improve ggml_vk_create_buffer error handling (#9898)

FanShupei committed 204 days ago

Verified 418f5eef

readme : update hot topics

ggerganov committed 204 days ago

Verified ba6f62eb

server : fix smart selection of available slot (#10120)

sasha0552 committed 204 days ago

Verified d865d147

ggml : remove ggml_scratch (#10121)

ggerganov committed 204 days ago

Verified 1804adb0

ggerganov committed 204 days ago

Verified 815fe72a

ggml : alloc ggml_contexts on the heap (whisper/2525)

ggerganov committed 204 days ago

Verified f221d562

build: fix build error in Windows env with OneAPI setup (#10107)

kylo5aby committed 205 days ago

Verified e597e507

llama : improve output buffer type selection (#10098)

slaren committed 205 days ago

Verified 85679d37

quantize : fix --keep-split (#10114)

slaren committed 205 days ago

Verified 1e9f9499

llama : fix buffer checks for mamba and rwk (#10111)

slaren committed 205 days ago

Verified c02e5ab2

loader: refactor tensor weights storage (#9935)

kylo5aby committed 205 days ago

Verified ab3d71f9

server : include scheme when printing URL (#10106)

bakkot committed 205 days ago

Verified 0a683e80

ggml : check tensor name lengths in gguf files (#10100)

slaren committed 205 days ago

Verified dea5e860

kompute: add mul_mat_q4_k shader (#10097)

slp committed 205 days ago

Verified 1329c0a7

kompute: add backend registry / device interfaces (#10045)

slp committed 206 days ago

Verified 61408e7f

ggml : fix memory leaks when loading invalid gguf files (#10094)

slaren committed 206 days ago

Verified b9e02e81

readme : more lora detail in main example readme (#10064)

richdougherty committed 206 days ago

Verified 6763f713

convert : more detailed convert lora usage docs (#10065)

richdougherty committed 206 days ago

Verified 79a2bc04

ggml : add Q4_0_8_8 RISC-V GEMV and GEMM kernels (#10029)

xctan committed 206 days ago

Verified fc83a9e5

llama : refactor model loader with backend registry (#10026)

slaren committed 207 days ago

Verified c5b0f4b5

ggml: Add POOL2D OP for GPU acceleration to the Vulkan backend in the MobileVLM model. (#9763)

cyzero-kim committed 207 days ago

Verified 8f275a7c

llama : remove Tail-Free sampling (#10071)

ggerganov committed 207 days ago

Verified 8d8ff715

llama : Add IBM granite template (#10013)

arch-btw committed 208 days ago

Verified 61715d5c

flake.lock: Update (#10063)

ggerganov committed 208 days ago

Verified 07028f9d

Older