summaryrefslogtreecommitdiff
path: root/candle-nn
Commit message (Expand)AuthorAgeFilesLines
* Sync upstream MLX sdpa vector kernels with mask (#2718)HEADmainEric Buehler2025-01-161-21/+74
* ModernBERT model (#2713)Jani Monoses2025-01-132-1/+12
* Clippy fixes for 1.84. (#2710)Laurent Mazare2025-01-101-2/+2
* Lint fixes introduced with Rust 1.83 (#2646)Anubhab Bandyopadhyay2024-11-282-10/+10
* Provide a method to allow PTH files with state maps to be loaded. (#2639)zachcp2024-11-261-1/+11
* Update docs (#2553)zachcp2024-11-119-0/+34
* Add some fast Metal MLX SDPA kernels (#2584)Eric Buehler2024-11-052-0/+396
* Improved launch config for layer-norm/rms-norm. (#2591)Laurent Mazare2024-11-042-4/+66
* Make the RNN configs accessible from the models. (#2541)Laurent Mazare2024-10-041-72/+103
* Add/lstm direction (#2455)Justin Sing2024-09-301-8/+25
* Add Pixtral. (#2521)Laurent Mazare2024-09-301-14/+22
* Add a RotatingKVCache. (#2493)Laurent Mazare2024-09-232-1/+333
* onnx: implement LSTM op (#2268)shua2024-08-191-0/+4
* update: LSTMState and GRUState fields to be public (#2384)Justin Sing2024-08-011-3/+3
* Add support for Llama 3.1 (#2359)Eric Buehler2024-07-262-3/+4
* Depth Anything v2 (#2279)Jeroen Vlek2024-06-241-1/+22
* add where_cond f32 for metal (#2236)Lionel Touati2024-06-021-1/+1
* Enable the new layer-norm. (#2213)Laurent Mazare2024-05-242-5/+19
* Add the layernorm specialized op. (#2212)Laurent Mazare2024-05-242-5/+280
* Simplify the KvCache api. (#2207)Laurent Mazare2024-05-231-36/+53
* Add a couple kv-cache helper functions. (#2206)Laurent Mazare2024-05-231-0/+29
* Add a slice_set op. (#2193)Laurent Mazare2024-05-182-0/+102
* Fix VarBuilder::from_slice_safetensors (#2180)Harry Stern2024-05-121-4/+30
* Add SliceSafetensors. (#2179)Laurent Mazare2024-05-111-0/+6
* Bump the version number to 0.5.1. (#2155)Laurent Mazare2024-05-031-1/+1
* Bug Fix: When converting a tensor to a variable, clone if the tensor is alrea...Jeffrey Dallatezza2024-04-291-2/+44
* Fix sigmoid gradient calculation and move sigmoid into a specialized op (#2114)MilkFather2024-04-292-2/+197
* Apply the cast before the scaling. (#2135)Laurent Mazare2024-04-281-1/+1
* Use the faster rms-norm kernel for llama. (#2107)Laurent Mazare2024-04-221-4/+13
* Add a convenient way to rename tensors accessed through a varbuilder. (#2052)Laurent Mazare2024-04-131-0/+93
* Add the rope THD kernel. (#2014)Laurent Mazare2024-04-052-0/+262
* Relax the contiguous check for cuda kernels. (#2000)Laurent Mazare2024-04-031-1/+1
* Add benchmarks for the candle-nn package (#1995)Thomas Santerre2024-04-035-0/+175
* Add fn 'get_with_hints_dtype' in VarBuilder (#1877) (#1897)yinqiwen2024-04-011-4/+15
* Fix detail in new RoPE implementation (#1935)Hugo Abonizio2024-03-251-1/+1
* Contiguous variant of the rope kernel. (#1929)Laurent Mazare2024-03-252-2/+282
* Fast kernels for rotary embeddings. (#1928)Laurent Mazare2024-03-244-0/+277
* RmsNorm kernel for metal. (#1895)Laurent Mazare2024-03-211-1/+46
* Custom op for RmsNorm (#1890)Laurent Mazare2024-03-212-8/+197
* Optimize the cat operation on contiguous tensors (#1855)Laurent Mazare2024-03-171-0/+19
* add clone to candle dropout (#1814)Kirpal Grewal2024-03-081-1/+1
* Improve metal buffer usage (#1807)ivarflakstad2024-03-071-1/+2
* Add the StarCoder2 model. (#1779)Laurent Mazare2024-02-281-0/+4
* Encodec model. (#1771)Laurent Mazare2024-02-271-1/+1
* Tweak the VarMap set type. (#1758)Laurent Mazare2024-02-252-2/+39
* Support for attention bias in gemma + refactor things a bit. (#1744)Laurent Mazare2024-02-222-6/+19
* Bugfix for applying the bias in conv1d-transpose. (#1732)Laurent Mazare2024-02-181-1/+1
* Support for groups in conv-transpose1d. (#1731)Laurent Mazare2024-02-181-3/+13
* Expose the weights and biases in transposed convolutions. (#1727)Laurent Mazare2024-02-181-0/+16
* Expose more conv1d functions/structs. (#1726)Laurent Mazare2024-02-172-2/+19
generated by cgit v1.2.3 (git 2.39.1) at 2025-04-02 17:15:59 +0000