Commit History

Автор SHA1 Съобщение Дата
  Alex Cheema aa7cd34b51 Merge pull request #323 from lipere123/adalovace преди 9 месеца
  lipere123 6c302244e0 Merge branch 'exo-explore:main' into adalovace преди 9 месеца
  Alex Cheema 9a93dcc414 Merge pull request #329 from varshith15/llava_broken_fix преди 9 месеца
  Varshith 0730532351 fix: tokenize преди 9 месеца
  Alex Cheema 0f272f53b8 Merge pull request #326 from exo-explore/fix_llama_embed_tokens преди 9 месеца
  Alex Cheema 17065d879b dynamic halfway partition point in unit test преди 9 месеца
  Alex Cheema ae74d2da16 run unit test on llama 3.2 1b for faster test преди 9 месеца
  Alex Cheema ad09b4b3d9 also initialize embed_tokens if last layer and tie_word_embeddings true преди 9 месеца
  Alex Cheema fbc407c6ae make llama-3.2-1b the default for tests so they run faster преди 9 месеца
  Alex Cheema 8950d95ef8 updgrade all mac ci jobs to xcode=16.0.0, resource_class=m2pro.large преди 9 месеца
  LIPERE Benjamin ade9db4dfd feat(device_capabilities.py): add support for NVIDIA RTX 4000 ADA generation device capabilities преди 9 месеца
  Alex Cheema c1a26cd7fa Merge pull request #315 from GaetanLepage/a6000-flops преди 9 месеца
  Gaetan Lepage a0ad18c6ad Fix GPU names for RTX Ampere cards преди 9 месеца
  Alex Cheema eade4fb62d Merge pull request #310 from exo-explore/udp_discovry_peer_race_condition преди 9 месеца
  Alex Cheema 8a69a7a238 one line print преди 9 месеца
  Alex Cheema b7996b9af4 race condition in on_listen_message for udp discovery fixes #308 преди 9 месеца
  Alex Cheema 27e5ee6b22 Merge pull request #309 from exo-explore/cleanup_peers_race_condition преди 9 месеца
  Alex Cheema e80ee60760 fix the race condition in cleanup peers and run the peer checks concurrently. fixes #308 преди 9 месеца
  Alex Cheema 48fe7b9130 Merge pull request #306 from exo-explore/process_prompt_shield преди 9 месеца
  Alex Cheema aa20562614 shield process_prompt so downloads dont get cancelled when chatgpt api request times out преди 9 месеца
  Alex Cheema 7b2a523fd1 Merge pull request #289 from GaetanLepage/pynvml преди 10 месеца
  Alex Cheema db123884e9 Merge pull request #291 from vanakema/main преди 10 месеца
  Alex Cheema 33c0dd8583 Merge pull request #292 from exo-explore/remove_tailscale_package_dependency преди 10 месеца
  Alex Cheema e8a8702377 replace tailscale.devices with good old http, removing the need for tailscale dependency преди 10 месеца
  Mark Van Aken 82c7ce69f6 Point `llama-3.1-70b-bf16` model to the actually bf16 version преди 10 месеца
  Gaetan Lepage 9ffd8116f0 Use official nvidia-ml-py instead of pynvml преди 10 месеца
  Alex Cheema 2b9dec20eb Merge pull request #282 from exo-explore/unsilence_errors преди 10 месеца
  Alex Cheema 9b9f40d470 only stream results for the same request id. this allows multiple concurrent requests on the same LLM without overlapping interference in the streamed outputs преди 10 месеца
  Alex Cheema 9223993e29 await node process_prompt with timeoout преди 10 месеца
  Alex Cheema 5e0db20426 Merge pull request #278 from exo-explore/peer_prio преди 10 месеца