 Georgi Gerganov
|
447ccbe8c3
readme : add new roadmap + manifesto
|
2 years ago |
|
Georgi Gerganov
|
bd34cdde38
ggml : sync latest ggml (custom operators)
|
2 years ago |
 anon998
|
c2a08f87b8
fix server sampling: top k sampler first (#1977)
|
2 years ago |
|
Georgi Gerganov
|
66a2555ba6
readme : add Azure CI discussion link
|
2 years ago |
 sjinzh
|
e65ca7e14a
zig : upgrade build system support (#1981)
|
2 years ago |
 Robyn
|
5ec8dd5a3c
#1869 Fix null reference errors when training from scratch with CUDA (#1907)
|
2 years ago |
|
Georgi Gerganov
|
65bdd52a86
tests : sync test-grad0 from ggml
|
2 years ago |
 Rowan Hart
|
fdd1860911
flake : fix ggml-metal.metal path and run nixfmt (#1974)
|
2 years ago |
 AN Long
|
c943d823c1
convert : fix invalid params in write_vocab_only (#1975)
|
2 years ago |
 slaren
|
f2c754e1c3
ggml : improve ggml_graph_dump_dot, add ggml_format_name (#1978)
|
2 years ago |
|
Georgi Gerganov
|
11da1a85cd
readme : fix whitespaces
|
2 years ago |
 Alberto
|
235b610d65
readme : fixed termux instructions (#1973)
|
2 years ago |
 Alex Renda
|
b061ba9e2a
llama : fix top-p sampling to match the canonical definition (#1953)
|
2 years ago |
 Didzis Gosko
|
527b6fba1d
llama : make model stateless and context stateful (llama_state) (#1797)
|
2 years ago |
 eiery
|
d7b7484f74
Add OpenLLaMA instructions to the README (#1954)
|
2 years ago |
 Erik Scholz
|
7487137227
rework convert.py to read hyper-parameters from config.json (#1958)
|
2 years ago |
 Johannes Gäßler
|
bbca06e269
cmake: revert CUDA arch default to 52, 61 if f16 (#1959)
|
2 years ago |
 Rahul Vivek Nair
|
fb98254f99
Fix typo in README.md (#1961)
|
2 years ago |
|
Georgi Gerganov
|
049aa16b8c
readme : add link to p1
|
2 years ago |
 Xiake Sun
|
2322ec223a
Fix typo (#1949)
|
2 years ago |
 Ettore Di Giacinto
|
aacdbd4056
llama : fix params struct slignment (#1936)
|
2 years ago |
 Henri Vasserman
|
20568fe60f
[Fix] Reenable server embedding endpoint (#1937)
|
2 years ago |
|
Georgi Gerganov
|
18b35625c3
ggml : fix bug in LBFGS optimizer (found by ggml tests)
|
2 years ago |
 l3utterfly
|
ba4e85a833
llama : use aligned memory during ggml_init call from loading saved sessions (#1934)
|
2 years ago |
|
Georgi Gerganov
|
23fc5c219a
cmake : fix trailing whitespaces
|
2 years ago |
 Kawrakow
|
cb40dfca69
llama : only use Q6_K for output weights if tensor size is multiple of 256 (#1932)
|
2 years ago |
|
Kawrakow
|
ca7c3f4da5
cuda : faster k-quants on older GPUs (#1930)
|
2 years ago |
|
Georgi Gerganov
|
b97ca431db
ggml : sync latest ggml repo (#1924)
|
2 years ago |
 Howard Su
|
1e3abfcef0
cmake : fix build shared ggml when CUDA is enabled (#1929)
|
2 years ago |
|
Johannes Gäßler
|
16b9cd1939
Convert vector to f16 for dequantize mul mat vec (#1913)
|
2 years ago |
