 Alex Cheema
|
eafed8e16e
fix legacy model loading
|
hai 1 ano |
|
Alex Cheema
|
08e8cacf01
run integration test for each inference engine
|
hai 1 ano |
|
Alex Cheema
|
32bb44b3a2
request to both nodes in integration test, dont preload the model - exo should be robust against that
|
hai 1 ano |
|
Alex Cheema
|
c06124c637
fix circleci badge
|
hai 1 ano |
 Alex Cheema
|
0b06516af8
Merge pull request #115 from exo-explore/circleci-project-setup
|
hai 1 ano |
|
Alex Cheema
|
f4b0f1ca02
replace tests badge with circleci
|
hai 1 ano |
|
Alex Cheema
|
e3a524fd89
Merge pull request #114 from exo-explore/circleci-project-setup
|
hai 1 ano |
|
Alex Cheema
|
96969e3540
prefix matching logic
|
hai 1 ano |
|
Alex Cheema
|
7c0923279c
trigger circleci
|
hai 1 ano |
|
Alex Cheema
|
2a8f1ae4c5
get rid of caches. circeci downloads fast enough
|
hai 1 ano |
|
Alex Cheema
|
54ea5dbb5c
only remove the matching prefix from the prompt if its length is less than the prompt
|
hai 1 ano |
|
Alex Cheema
|
67ad3f57a1
use llama 3.1 in tests
|
hai 1 ano |
|
Alex Cheema
|
92b66e275f
fix cache
|
hai 1 ano |
|
Alex Cheema
|
cf3dae9570
circleci: separate hf, tinygrad caches
|
hai 1 ano |
|
Alex Cheema
|
faadfa29dd
circleci chatgpt integration test
|
hai 1 ano |
|
Alex Cheema
|
9a03991f1a
cache key
|
hai 1 ano |
|
Alex Cheema
|
2065c7751a
orb
|
hai 1 ano |
|
Alex Cheema
|
cf4cddccc1
port github workflow to circleci
|
hai 1 ano |
|
Alex Cheema
|
21b0bf8777
Update config.yml
|
hai 1 ano |
|
Alex Cheema
|
997fcaff35
CircleCI Commit
|
hai 1 ano |
|
Alex Cheema
|
f5755ea198
give explicit node ids when running on the same instance in tests, otherewise they use the same one because of sticky node ids
|
hai 1 ano |
|
Alex Cheema
|
4faa6c06f4
add support for selective model downloading. related: #16
|
hai 1 ano |
|
Alex Cheema
|
be3a09c28b
rm logs
|
hai 1 ano |
|
Alex Cheema
|
0bfb8e3b6d
sticky node ids #16
|
hai 1 ano |
|
Alex Cheema
|
980d5d2c5b
bring back TINYGRAD_DEBUG. not sure why it was removed
|
hai 1 ano |
|
Alex Cheema
|
76766253cd
fix regression introduced by image_str for tinygrad
|
hai 1 ano |
|
Alex Cheema
|
1d54f10514
pass on tinygrad set_on_download_progress
|
hai 1 ano |
|
Alex Cheema
|
d6a7e46324
async model downloading with download progress. fixes #102. related: #16 #104
|
hai 1 ano |
|
Alex Cheema
|
5c67e24c35
smart prompt longest prefix matching to avoid sending the same text through the NN again. speeds up prefill significantly
|
hai 1 ano |
|
Alex Cheema
|
94ac9463a7
fix model id for llama 3.1 405b now its finally on the hub
|
hai 1 ano |
