Alex Cheema
|
355c579965
more robust discovery / peer handling. now we track if the same node id changes address, then we immediately conenct to it
|
11 months ago |
Alex Cheema
|
8114a79ede
add back listen and cleanup tasks
|
11 months ago |
Alex Cheema
|
dcb3ac76a8
test kill pids
|
11 months ago |
Alex Cheema
|
3dd81a1e05
fix UDPDiscovery params, create a new transport every time we broadcast
|
11 months ago |
Alex Cheema
|
15b5043d6e
test for reconnect
|
11 months ago |
Alex Cheema
|
baf6efd321
cleaner discovery
|
11 months ago |
Alex Cheema
|
8baaad7f6b
Merge branch 'main' into better_networking
|
11 months ago |
Alex Cheema
|
57215041a0
todo for speculative model
|
11 months ago |
Alex Cheema
|
dc3b2bde39
use NousResearch/Meta-Llama-3.1-70B-Instruct as tinygrad llama-3.1-70b model, previously using non-instruct model
|
11 months ago |
Alex Cheema
|
12609cb6e4
integration test for udp discovery with grpc server
|
11 months ago |
Alex Cheema
|
f93f811dcb
generalise UDPDiscovery to any kind of PeerHandle that accepts an address. test it
|
11 months ago |
Alex Cheema
|
d4a932e405
fix merge
|
11 months ago |
Alex Cheema
|
2341aa1acf
Merge branch 'main' into better_networking
|
11 months ago |
Alex Cheema
|
5a9f4ba5c1
update examples: remove old llama3_distributed, add chatgpt_api
|
11 months ago |
Alex Cheema
|
581856897a
clean up unused, formatting
|
11 months ago |
Alex Cheema
|
62e3726263
add RTX 20 series to device capabilities
|
11 months ago |
Alex Cheema
|
ebff636a25
script ot start openwebui
|
11 months ago |
Alex Cheema
|
394935711b
add all chat endpoints without v1 prefix to support ollama / openwebui. related: #175
|
11 months ago |
Alex Cheema
|
70172d7cb9
add /v1/models endpoint and change Content-Type of stremed response to text/event-stream. fixes #175
|
11 months ago |
Alex Cheema
|
d917778e2b
update mlx to 0.17.1 (not sure where 0.17.0 went on PyPi disappeared)g
|
11 months ago |
Alex Cheema
|
2667c8af44
cleaner download_progress
|
11 months ago |
Alex Cheema
|
f46d077beb
fix font dependencies for tinychat. related: #172
|
11 months ago |
Alex Cheema
|
8a4928f80c
fix gitignore to not ignore tinychat static files
|
11 months ago |
Alex Cheema
|
d515d9efa3
explicitly use absolute paths for tinychat deps
|
11 months ago |
Alex Cheema
|
a386c35fde
script to update tinychat deps
|
11 months ago |
Alex Cheema
|
3791e669a4
download tinychat dependencies all to local dir so we dont need internet
|
11 months ago |
Alex Cheema
|
85bab25ac0
fix local check if dir does not exist
|
11 months ago |
Alex Cheema
|
59c4393d95
first try loading tokenizer from local path instead of always going to the internet first. significant speed ups
|
11 months ago |
Alex Cheema
|
784e6bae21
print traceback on topology collection error
|
11 months ago |
Alex Cheema
|
8cad0e1849
only use_fast tokenizer for Mistral Large until this inconsistency bug is fixed #171
|
11 months ago |