Commit Graph

  • 66750f9827 Add some 'cuda-if-available' helper function. (#172) Laurent Mazare 2023-07-15 08:25:15 +01:00
  • 2ddda706bd Switch to using trunk. (#171) Laurent Mazare 2023-07-14 22:06:40 +01:00
  • d1f5d44c04 Reenable pyo3 in the workspace list (#170) Laurent Mazare 2023-07-14 19:54:38 +01:00
  • d1f6fad84a Whisper example in wasm. (#169) Laurent Mazare 2023-07-14 19:33:36 +01:00
  • 3f930bdaee Merge pull request #168 from LaurentMazare/remove_cuda_default Nicolas Patry 2023-07-14 18:08:52 +02:00
  • 3672e1a46f Revert "Testing fmt CI check behind cuda feature flag." Nicolas Patry 2023-07-14 15:18:14 +00:00
  • b9605310b1 Testing fmt CI check behind cuda feature flag. Nicolas Patry 2023-07-14 15:14:52 +00:00
  • dcb4a9291e Expliciting how to enable cuda. Nicolas Patry 2023-07-14 17:08:05 +02:00
  • 2d5e952cf9 Updating the example to reflect new command lines. Nicolas Patry 2023-07-14 16:56:33 +02:00
  • 4ed56d7861 Removing cuda default. Nicolas Patry 2023-07-14 16:52:15 +02:00
  • 88f666781f Wasm proof of concept. (#167) Laurent Mazare 2023-07-14 14:51:46 +01:00
  • d88b6cdca9 Add backtrace information to errors where relevant. (#166) Laurent Mazare 2023-07-14 09:31:25 +01:00
  • a2f72edc0d Simplify the parameters used by sum and sum_keepdim. (#165) Laurent Mazare 2023-07-14 08:22:08 +01:00
  • 2bfa791336 Use the same default as pytorch for sum. (#164) Laurent Mazare 2023-07-13 21:32:32 +01:00
  • 57be3638d8 Add the pytorch version of the linear regression as a comment. (#163) Laurent Mazare 2023-07-13 21:05:57 +01:00
  • 23e105cd94 Add the gradient for reduce-sum. (#162) Laurent Mazare 2023-07-13 20:14:10 +01:00
  • 3c02ea56b0 Add a cli argument to easily switch the dtype. (#161) Laurent Mazare 2023-07-13 19:18:49 +01:00
  • ded93a1169 Add the SGD optimizer (#160) Laurent Mazare 2023-07-13 19:05:44 +01:00
  • 5ee3c95582 Move the variable creation to the variable module. (#159) Laurent Mazare 2023-07-13 16:55:40 +01:00
  • 6991036bc5 Introduce the variables api used for adjusting parameters during the training loop. (#158) Laurent Mazare 2023-07-13 14:09:51 +01:00
  • 7adc8c903a Expose the storage publicly. (#157) Laurent Mazare 2023-07-13 13:52:36 +01:00
  • 21aa29ddce Use a rwlock for inner mutability. (#156) Laurent Mazare 2023-07-13 11:25:24 +01:00
  • dfabc708f2 Fix a comment. (#155) Laurent Mazare 2023-07-13 11:11:37 +01:00
  • 50b0946a2d Tensor mutability (#154) Laurent Mazare 2023-07-13 11:04:40 +01:00
  • a3663ce2f2 Encodec forward pass (#153) Laurent Mazare 2023-07-13 08:18:39 +01:00
  • 6c75a98ad2 Add the forward pass for the T5 model. (#152) Laurent Mazare 2023-07-12 22:02:40 +01:00
  • 465fc8c0c5 Add some documentation and test to the linear layer. (#151) Laurent Mazare 2023-07-12 20:24:23 +01:00
  • f09d7e5653 Merge pull request #150 from LaurentMazare/add_cheatsheet Nicolas Patry 2023-07-12 19:54:07 +02:00
  • 91817b9e57 Tweak the table formatting. laurent 2023-07-12 18:43:52 +01:00
  • a86ec4b9f0 Add more documentation and examples. (#149) Laurent Mazare 2023-07-12 17:40:17 +01:00
  • 6e938cfe9d Adding cheatsheet + expand on other ML frameworks. Nicolas Patry 2023-07-12 18:35:34 +02:00
  • 8aab787384 Test the index op + bugfix. (#148) Laurent Mazare 2023-07-12 15:42:36 +01:00
  • ba35d895e7 Sketch the candle-transformers crate. (#147) Laurent Mazare 2023-07-12 13:49:31 +01:00
  • eae646d322 Use arange in the examples. (#146) Laurent Mazare 2023-07-12 12:12:34 +01:00
  • 20599172ac Add from_iter and arange, use it in the doctests. (#145) Laurent Mazare 2023-07-12 12:03:01 +01:00
  • b3b39cca92 Llama batch (#144) Laurent Mazare 2023-07-12 11:38:19 +01:00
  • bcf96e3cf3 Implement the backend trait for the cpu backend. (#143) Laurent Mazare 2023-07-12 09:54:33 +01:00
  • a76ec797da Cleanup the main crate error and add a couple dedicated ones (#142) Laurent Mazare 2023-07-12 09:17:08 +01:00
  • fa760759e5 Allow for lazy loading of npz files, use it in llama to reduce memory usage in the cpu version. (#141) Laurent Mazare 2023-07-11 20:22:34 +01:00
  • 37cad85869 Resurrect the llama npy support. (#140) Laurent Mazare 2023-07-11 19:32:10 +01:00
  • 760f1d7055 Refactor the llama example to make it more in sync with the other ones. (#139) Laurent Mazare 2023-07-11 17:20:55 +01:00
  • 64264d97c1 Modular backends (#138) Laurent Mazare 2023-07-11 11:17:02 +01:00
  • 674eb35e10 Remove some dead-code pragmas. (#137) Laurent Mazare 2023-07-11 09:33:59 +01:00
  • ae79c00e48 Allow for uniform initialization in a single step. (#136) Laurent Mazare 2023-07-11 08:52:29 +01:00
  • b31a3bbdcb Sketch the tensor initialization module. (#134) Laurent Mazare 2023-07-11 07:41:46 +01:00
  • 0e9d3afd77 Simplify the var-builder layer setup. (#133) Laurent Mazare 2023-07-10 23:22:58 +01:00
  • 6fc1ab4f0d MusicGen var-store path cleanup. (#132) Laurent Mazare 2023-07-10 23:13:11 +01:00
  • b46c28a2ac VarBuilder path creation (#131) Laurent Mazare 2023-07-10 22:37:34 +01:00
  • 1aa7fbbc33 Move the var-builder in a central place. (#130) Laurent Mazare 2023-07-10 20:49:50 +01:00
  • 2be09dbb1d Macroify the repeating bits. (#129) Laurent Mazare 2023-07-10 19:44:06 +01:00
  • 23849cb6e6 Merge pull request #124 from LaurentMazare/new_doc Nicolas Patry 2023-07-10 20:43:23 +02:00
  • fba07d6b6b Merge pull request #127 from LaurentMazare/tensor_indexing Nicolas Patry 2023-07-10 19:56:34 +02:00
  • 1ad235953b Clippy ? Nicolas Patry 2023-07-10 19:34:38 +02:00
  • c9d354f5ae Update candle-core/src/tensor.rs Nicolas Patry 2023-07-10 19:29:22 +02:00
  • f29b77ec19 Random initializers. (#128) Laurent Mazare 2023-07-10 18:26:21 +01:00
  • 5ea747c047 Update candle-core/src/indexer.rs Nicolas Patry 2023-07-10 19:02:35 +02:00
  • ef0375d8bc i(..) indexing sugar (partial). Nicolas Patry 2023-07-10 17:34:04 +02:00
  • e2807c78a4 Enable the doctests to run with mkl (though they are broken for now). (#126) Laurent Mazare 2023-07-10 16:27:46 +01:00
  • 548b1df7ea Remove the dependency to blas and use mkl directly. (#125) Laurent Mazare 2023-07-10 15:52:03 +01:00
  • e01d099b71 Squeeze/unsqueeze/reshape Nicolas Patry 2023-07-10 16:40:25 +02:00
  • 221b1aff65 Support dgemm in mkl matmul. (#122) Laurent Mazare 2023-07-10 15:02:37 +01:00
  • 71cd3745a9 Add some layer-norm tests. (#121) Laurent Mazare 2023-07-10 14:43:04 +01:00
  • dc58259679 Merge pull request #120 from LaurentMazare/some_doc_plus_fix_stack Nicolas Patry 2023-07-10 15:21:24 +02:00
  • 9a667155fd Removed commented deny Nicolas Patry 2023-07-10 15:18:23 +02:00
  • 2c8fbe8155 oops. Nicolas Patry 2023-07-10 15:13:52 +02:00
  • 49f4a77ffd Put them back. Nicolas Patry 2023-07-10 15:11:48 +02:00
  • 38ac50eeda Adding some doc + Extended stack to work with extra final dimensions. Nicolas Patry 2023-07-10 14:51:10 +02:00
  • 204618b7d3 Merge pull request #118 from LaurentMazare/readme_update Nicolas Patry 2023-07-10 13:12:23 +02:00
  • 868743b8b9 Expanding a bit the README Nicolas Patry 2023-07-10 12:51:37 +02:00
  • 89a5b602a6 Move the conv1d layer to candle_nn. (#117) Laurent Mazare 2023-07-10 11:02:06 +01:00
  • b06e1a7e54 [nn] Move the Embedding and Activation parts. (#116) Laurent Mazare 2023-07-10 10:24:52 +01:00
  • 9ce0f1c010 Sketch the candle-nn crate. (#115) Laurent Mazare 2023-07-10 08:50:09 +01:00
  • bc3be6f9b0 Add the elu cuda kernel. (#114) Laurent Mazare 2023-07-10 07:57:01 +01:00
  • 270997a055 Add the elu op. (#113) Laurent Mazare 2023-07-09 21:56:31 +01:00
  • ea5dfa69bc Sketching the musicgen model. (#66) Laurent Mazare 2023-07-09 19:53:35 +01:00
  • c187f347bf Make it easier to use whisper samples from the repo. (#112) Laurent Mazare 2023-07-08 18:48:27 +01:00
  • eb64ad0d4d Cuda kernel for the conv1d op (#111) Laurent Mazare 2023-07-08 18:13:25 +01:00
  • 5c3864f9f7 Add more sum tests. (#110) Laurent Mazare 2023-07-08 13:15:36 +01:00
  • e676f85f00 Sketch a fast cuda kernel for reduce-sum. (#109) Laurent Mazare 2023-07-08 12:43:56 +01:00
  • 33479c5f1b Add some very simple sum benchmark. (#108) Laurent Mazare 2023-07-08 08:39:27 +01:00
  • f35cfc5e97 Sample with temperature. (#106) Laurent Mazare 2023-07-07 18:12:25 +01:00
  • 03dffe9ecc Use F32 for the reduce ops. (#105) Laurent Mazare 2023-07-07 17:55:21 +01:00
  • e923b3adc2 Add a KV cache to falcon. (#104) Laurent Mazare 2023-07-07 17:24:38 +01:00
  • 05ff1cff66 Add some caching to the causal mask. (#103) Laurent Mazare 2023-07-07 12:56:44 +01:00
  • 65937612d0 Merge pull request #91 from LaurentMazare/tweak_parallel_download Nicolas Patry 2023-07-07 09:43:55 +02:00
  • 2df044f9a1 Clippy after rebase. Nicolas Patry 2023-07-07 09:22:09 +02:00
  • 1ec221a749 Fixing falcon example. Nicolas Patry 2023-07-07 00:29:44 +02:00
  • 514b171f75 Getting tokio tasks stuck on smaller machines. Nicolas Patry 2023-07-06 14:21:35 +02:00
  • d38a926c14 Convert the logits to f32 before extracting them. (#102) Laurent Mazare 2023-07-07 08:07:57 +01:00
  • 02b5c38049 Use cublas bf16. (#101) Laurent Mazare 2023-07-07 08:00:12 +01:00
  • c71a38deb7 Tweak the include order to include math.h first. (#100) Laurent Mazare 2023-07-07 06:47:25 +01:00
  • f114394456 Include the math.h file to get access to constants. (#99) Laurent Mazare 2023-07-07 06:42:57 +01:00
  • bac4ef40f3 Add some text generation pipeline for falcon. (#98) Laurent Mazare 2023-07-07 06:34:22 +01:00
  • 2b8e8c9f14 Bugfixes. (#97) Laurent Mazare 2023-07-06 23:26:11 +01:00
  • a3f3b93d16 Add the call to dense in the attention layer. (#96) Laurent Mazare 2023-07-06 23:22:08 +01:00
  • 0a2c82e301 Merge pull request #92 from LaurentMazare/sync_hub Nicolas Patry 2023-07-07 00:10:47 +02:00
  • 666c6f07a1 Merge pull request #88 from LaurentMazare/fix_unsafe_loads Nicolas Patry 2023-07-07 00:09:56 +02:00
  • ce27073feb Update candle-core/src/safetensors.rs Nicolas Patry 2023-07-06 23:59:54 +02:00
  • 0f679fe42e Fix some shape issues in falcon. (#95) Laurent Mazare 2023-07-06 19:23:54 +01:00
  • 4afa461b34 Sketch the Falcon model. (#93) Laurent Mazare 2023-07-06 19:01:21 +01:00