 Alex Cheema
|
5c67e24c35
smart prompt longest prefix matching to avoid sending the same text through the NN again. speeds up prefill significantly
|
il y a 1 an |
|
Alex Cheema
|
94ac9463a7
fix model id for llama 3.1 405b now its finally on the hub
|
il y a 1 an |
|
Alex Cheema
|
178fb75c84
fix image api prompt encoding
|
il y a 1 an |
|
Alex Cheema
|
2d20000964
use AutoProcessor with use_fast=False since there's a bug with use_fast=True where whitespace is removed on single token decodes
|
il y a 1 an |
 Alex Cheema
|
0ec77e1a99
Merge pull request #88 from varshith15/main
|
il y a 1 an |
|
Alex Cheema
|
af1c7ce327
add support for image upload to tinychat for vision models
|
il y a 1 an |
|
Alex Cheema
|
0d45a855fb
increase max request size to send raw images, make image download from url async, use chatgpt-compatible convention for images
|
il y a 1 an |
|
Alex Cheema
|
e68d06f4ef
move model-selector styles to index.css
|
il y a 1 an |
|
Alex Cheema
|
78db451d7e
add pillow to main dependencies
|
il y a 1 an |
|
Alex Cheema
|
824f05263f
Merge branch 'main' into HEAD
|
il y a 1 an |
|
Alex Cheema
|
142682645f
bump up tinygrad version
|
il y a 1 an |
 Varshith
|
8d3d3df1dd
update readme
|
il y a 1 an |
|
Varshith
|
acc94b50c7
chatgpt api integration
|
il y a 1 an |
|
Alex Cheema
|
33cbacf513
fix llava sanitize
|
il y a 1 an |
|
Alex Cheema
|
2fb961fccd
stick to same convention as new llama
|
il y a 1 an |
|
Alex Cheema
|
b44b917151
add pillow as testing dependency
|
il y a 1 an |
|
Alex Cheema
|
2aa1e24ea9
remove unused torch import
|
il y a 1 an |
|
Alex Cheema
|
833e7f3396
rename sharded_llava -> llava to match new convention
|
il y a 1 an |
|
Alex Cheema
|
7d5eed1111
Merge branch 'main' into HEAD
|
il y a 1 an |
|
Alex Cheema
|
044d189ccc
Merge pull request #94 from mzbac/mlx_refactor
|
il y a 1 an |
|
Alex Cheema
|
909d5ef8ba
Merge branch 'main' into mlx_refactor
|
il y a 1 an |
|
Alex Cheema
|
63e51a8270
formatting
|
il y a 1 an |
|
Alex Cheema
|
6695b019a2
format format.py
|
il y a 1 an |
|
Alex Cheema
|
1dc08fecaa
increase max line length to 200
|
il y a 1 an |
|
Alex Cheema
|
444137776a
formatting
|
il y a 1 an |
 Anchen
|
a6bb8ddf41
update deepseek sanitize to shard layers first before handle switch
|
il y a 1 an |
|
Alex Cheema
|
cb217b7b77
format format.py
|
il y a 1 an |
|
Alex Cheema
|
4cb36a7f55
increase max line length to 200
|
il y a 1 an |
|
Alex Cheema
|
d94e3f9ce4
formatting
|
il y a 1 an |
|
Anchen
|
666b1c83ee
refactor(mlx): model sharding and add deepseek v2 support
|
il y a 1 an |
