summaryrefslogtreecommitdiff
path: root/candle-pyo3/src
Commit message (Collapse)AuthorAgeFilesLines
* Lint fixes introduced with Rust 1.83 (#2646)Anubhab Bandyopadhyay2024-11-281-1/+1
| | | | | | | | | | | * Fixes for lint errors introduced with Rust 1.83 * rustfmt * Fix more lints. --------- Co-authored-by: Laurent <laurent.mazare@gmail.com>
* pyo3 update. (#2545)Laurent Mazare2024-10-062-16/+15
| | | | | * pyo3 update. * Stub fix.
* Update for pyo3 0.21. (#1985)Laurent Mazare2024-04-012-45/+45
| | | | | | | | | | | * Update for pyo3 0.21. * Also adapt the RL example. * Fix for the pyo3-onnx bindings... * Print details on failures. * Revert pyi.
* Expose candle gather op in pyo3. (#1870)Laurent Mazare2024-03-181-0/+6
|
* Detach the tensors on batch-norm eval. (#1702)Laurent Mazare2024-02-131-2/+2
| | | | | | | | | | | | | * Detach the tensors on batch-norm eval. * Fix pyo3 bindings. * Black tweak. * Formatting. * Also update the pyo3-onnx formatting. * Apply black.
* Quantized GGUF style (#1523)Nicolas Patry2024-01-171-20/+31
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | * Metal quantized modifications proposal. - Add a device param, wherever needed. - Create new QMetal storage thing that implements QuantizedType. - Update everywhere needed. Fix Python. Fixing examples. Fix: fmt + clippy + stub. Moving everything around. Only missing the actual implems. Fixing everything + adding dequantized kernels. More work. Fixing matmul. Fmt + Clippy Some clippy fixes. Working state. Q2K Metal -> Bugged (also present in GGML). Q4K CPU -> Bugged (present previously, new test catch it). Q5K CPU -> Bugged (present previously). Q8_1 Both -> Never really implemented it seems Q8K metal -> Never implemented in metal Fixing Q2K bug (present in ggml). * Cleanup. * Fix the rebase. * Removing the fences speeds everything up and *is* correct this time... * Cleanup the fence. * After rebase. * Bad code removal. * Rebase after phi2 merge + fix replit default to CPU. * Making the CI happy. * More happy tests. --------- Co-authored-by: Nicolas Patry <nicolas@Nicolass-MacBook-Pro.local>
* Fix a couple typos (#1451)Laurent Mazare2023-12-171-9/+8
| | | | | * Mixtral quantized instruct. * Fix a couple typos.
* Implement the module trait directly for QMatMul. (#1372)Laurent Mazare2023-11-251-1/+1
|
* Metal part 1 - Scaffolding for metal. (#1308)Nicolas Patry2023-11-101-0/+13
| | | | | * Metal part 1 - Scaffolding for metal. * Remove tracing.
* PyO3: Add optional `candle.onnx` module (#1282)Lukas Kreussel2023-11-083-3/+237
| | | | | | | | | | | | | | | | | | | | | | | | | | | * Start onnx integration * Merge remote-tracking branch 'upstream/main' into feat/pyo3-onnx * Implement ONNXModel * `fmt` * add `onnx` flag to python ci * Pin `protoc` to `25.0` * Setup `protoc` in wheel builds * Build wheels with `onnx` * Install `protoc` in manylinux containers * `apt` -> `yum` * Download `protoc` via bash script * Back to `manylinux: auto` * Disable `onnx` builds for linux
* PyO3: Add `equal` and `__richcmp__` to `candle.Tensor` (#1099)Lukas Kreussel2023-10-301-2/+71
| | | | | | | | | | | | | | | | | | | * add `equal` to tensor * add `__richcmp__` support for tensors and scalars * typo * more typos * Add `abs` + `candle.testing` * remove duplicated `broadcast_shape_binary_op` * `candle.i16` => `candle.i64` * `tensor.nelements` -> `tensor.nelement` * Cleanup `abs`
* PyO3: Better shape handling (#1143)Lukas Kreussel2023-10-292-36/+134
| | | | | | | | | | | * Negative and `*args` shape handling * Rename to `PyShapeWithHole` + validate that only one hole exists * Regenerate stubs --------- Co-authored-by: Laurent Mazare <laurent.mazare@gmail.com>
* convert pytorch's tensor in Python API (#1172)andrew2023-10-251-0/+24
| | | | | * convert pytorch's tensor * separate tests for convert pytorch tensor
* Add support for accelerate in the pyo3 bindings. (#1167)Laurent Mazare2023-10-241-0/+3
|
* PyO3: Add `mkl` support (#1159)Lukas Kreussel2023-10-231-0/+3
| | | | | * Add `mkl` support * Set `mkl` path on linux
* PyO3: Add `None` and `Tensor` indexing to `candle.Tensor` (#1098)Lukas Kreussel2023-10-201-32/+94
| | | | | * Add proper `None` and `tensor` indexing * Allow indexing via lists + allow tensor/list indexing outside of first dimension
* PyO3: Add pytorch like `.to()` operator to `candle.Tensor` (#1100)Lukas Kreussel2023-10-191-0/+106
| | | | | * add `.to()` operator * Only allow each value to be provided once via `args` or `kwargs`
* Always broadcast magic methods (#1101)Lukas Kreussel2023-10-171-4/+4
|
* Add the pooling operators to the pyo3 layer. (#1086)Laurent Mazare2023-10-131-0/+24
|
* Make the Python Wrapper more Hackable and simplify Quantization (#1010)Lukas Kreussel2023-10-061-4/+130
| | | | | | | | | | | | | | | | | | | | | | | | | | | * Some first `Module` implementations * Add `state_dict` and `load_state_dict` functionality * Move modules around and create `candle.nn.Linear` * Add `nn.Embedding` and `nn.LayerNorm` * Add BERT implementation * Batch q-matmul * Automatically dequantize `QTensors` if a `Tensor` is expected * Add Module `.to()`, `.cuda()`, `cpu()` and `.type()` functionality * Unittests for `Module`, `Tensor` and `candle.utils` * Add `pytorch` like slicing to `Tensor` * Cleanup and BERT fixes * `black` formatting + unit-test for `nn.Linear` * Refactor slicing implementation
* Improve the quantized whisper setup. (#1018)Laurent Mazare2023-10-021-1/+1
| | | | | | | * Improve the quantized whisper setup. * Fix the config file paths. * Use the standard matmul where possible.
* Add return types to `*.pyi` stubs (#880)Lukas Kreussel2023-09-171-5/+265
| | | | | | | | | * Start generating return types * Finish tensor type hinting * Add `save_gguf` to `utils` * Typehint `quant-llama.py`
* Generate `*.pyi` stubs for PyO3 wrapper (#870)Lukas Kreussel2023-09-161-31/+58
| | | | | | | | | | | | | | | * Begin to generate typehints. * generate correct stubs * Correctly include stubs * Add comments and typhints to static functions * ensure candle-pyo3 directory * Make `llama.rope.freq_base` optional * `fmt`
* Add a python function to save as safetensors. (#740)Laurent Mazare2023-09-041-1/+13
|
* Return the metadata in the gguf pyo3 bindings. (#729)Laurent Mazare2023-09-041-4/+37
| | | | | | | * Return the metadata in the gguf pyo3 bindings. * Read the metadata in the quantized llama example. * Get inference to work on gguf files.
* Handle arbitrary shapes in Tensor::new. (#718)Laurent Mazare2023-09-021-5/+20
|
* More quantized llama in python. (#716)Laurent Mazare2023-09-021-5/+51
| | | | | | | | | * More quantized llama in python. * Expose a couple more functions. * Apply the last layer. * Use the vocab from the ggml files.
* Sketch a quantized llama using the pyo3 api. (#715)Laurent Mazare2023-09-021-6/+105
| | | | | | | | | | | * Sketch a quantized llama using the pyo3 api. * Add more ops. * Expose a few more functions to use in the quantized model. * Rope embeddings. * Get the forward pass to work.
* Add some quantized functions to pyo3. (#708)Laurent Mazare2023-09-011-1/+44
|
* Support for quantized tensors in the python api. (#706)Laurent Mazare2023-09-011-4/+155
| | | | | | | | | | | | | | | * Add more pyo3 support. * Add some support for quantized tensors in pyo3. * Add an arc layer on qmatmul. * Add the quantized matmul. * Quantization support. * More quantization support. * Test the python quantization.
* Cleanup the pyo3 setup. (#705)Laurent Mazare2023-09-011-0/+1
|
* Fixes for clippy 1.72. (#587)Laurent Mazare2023-08-241-0/+1
|
* Add support for i64 (#563)Laurent Mazare2023-08-231-0/+6
| | | | | * Add the i64 dtype. * Adapt the cuda kernels.
* Pyo3 dtype (#327)Laurent Mazare2023-08-061-19/+42
| | | | | * Better handling of dtypes in pyo3. * More pyo3 dtype.
* Add some tensor creation functions to the pyo3 bindings. (#326)Laurent Mazare2023-08-061-5/+69
|
* Use the same default as pytorch for sum. (#164)Laurent Mazare2023-07-131-2/+4
|
* Tensor mutability (#154)Laurent Mazare2023-07-131-1/+1
| | | | | * Working towards tensor mutability. * Use a ref-cell to provide tensor mutability.
* Get cuda to work on pyo3.laurent2023-07-021-0/+22
|
* Preliminary pyo3 support for device.laurent2023-07-021-2/+45
|
* Add the tensor function.laurent2023-07-021-0/+6
|
* Add dtype support.laurent2023-07-021-6/+29
|
* Add matmul/where_cond.laurent2023-07-021-0/+19
|
* Expose a couple more ops.laurent2023-07-021-5/+82
|
* Handle more input types to create tensors.laurent2023-07-021-2/+18
|
* Rename the method.laurent2023-07-021-1/+2
|
* Support higher order shapes for conversions.laurent2023-07-021-1/+31
|
* Add a trait to avoid repeating the dtype matching.laurent2023-07-021-20/+47
|
* More pyo3.laurent2023-07-011-2/+52
|
* Get shape to return a tuple.laurent2023-07-011-2/+3
|
* Add some binary operators.laurent2023-07-011-1/+17
|
generated by cgit v1.2.3 (git 2.39.1) at 2025-04-04 18:34:56 +0000