ngxson/llama.cpp

Pull Requests Commits

allow merging embd, output, etc

ngxson committed 1 year ago

cc89323f

implement slerp

ngxson committed 1 year ago

10c477b8

ngxson committed 1 year ago

a032bb6c

wip: new format

ngxson committed 1 year ago

65730438

ngxson committed 1 year ago

52186adc

ngxson committed 1 year ago

886e68ae

ngxson committed 1 year ago

853b3d71

merge: quant input quant output

ngxson committed 1 year ago

a1bf1e1b

merge: add debug msg

ngxson committed 1 year ago

f09188e9

merge: accept quant input

ngxson committed 1 year ago

abec8c0c

merge: try..catch

ngxson committed 1 year ago

2cfae6d9

merge: new input format

ngxson committed 1 year ago

3e6e3668

Merge branch 'master' into xsn/model_merge

ngxson committed 1 year ago

b6da762d

llama : constified `llama_set_state_data`'s `src` (#5774)

MarcusDunn committed 1 year ago

Verified d5ab2975

ci : reduce 3b ppl chunks to 1 to avoid timeout (#5771)

ggerganov committed 1 year ago

Verified 87c91c07

make portability_enumeration_ext apple only (#5757)

netrunnereve committed 1 year ago

Verified 317709b2

llama : remove deprecated API (#5770)

ggerganov committed 1 year ago

Verified 08c5ee87

awq-py : remove (#5768)

ggerganov committed 1 year ago

Verified 78aacf36

ngxson committed 1 year ago

f50bf00c

ggerganov committed 1 year ago

Verified 8c0e8f4e

add google magika inference example (ggml/748)

slaren committed 1 year ago

Verified 2774b0c9

Introduce backend GUIDs (ggml/743)

uextm committed 1 year ago

Verified 5f706718

server : hit Ctrl+C twice to exit (#5734)

ngxson committed 1 year ago

Verified a693bea1

llama : fix non-quantization of expert gating tensors (#5754)

compilade committed 1 year ago

Verified adcb12a9

llama : improve BERT tokenization (#5740)

iamlemec committed 1 year ago

Verified 177628bf

readme : add link to LLaVA 1.6 models (#5758)

danbev committed 1 year ago

Verified 6c441686

server : add "/chat/completions" alias for "/v1/...` (#5722)

jorgealias committed 1 year ago

Verified efc72253

ggml : make i-quants work with super-blocks of 64 (CPU,Metal) (#5760)

ikawrakow committed 1 year ago

Verified 7c4263d4

Attempt to fix android build (#5752)

ikawrakow committed 1 year ago

Verified cb49e0f8

IQ4_XS: a 4.25 bpw quantization (#5747)

ikawrakow committed 1 year ago

Verified 0becb22a

Older