82882cfd3eAdd Qwen3VL bf16 example implementation and tutorial docs, including BMP/JPG/PNG support via zignal library.
master
Foke Singh2025-12-31 12:46:11 +0000
e659dc8fa3Add Qwen3VL bf16 example implementation and integrate zignal image format support; update Bazel build files and core ZML modules.Tarry Singh2025-12-29 16:17:11 +0000
57bef8d66cSwitch workspace to use apple_support C++ toolchains instead of rules_cc, aligning with the recommended toolchain.Tarry Singh2025-12-22 17:33:18 +0000
6c80f2f394Set LLVM backend as default for all Zig compile actions via rules_zig zigopt global setting, due to stability concerns with the self-hosted backend.Tarry Singh2025-12-18 10:04:04 +0000
a3abf148b4Fix pickle loader to use takeDelimiterInclusive for Zig 0.15.2 and update ZLS runner, buffer, callback, and tracer utilities accordingly.Tarry Singh2025-12-15 11:08:19 +0000
1b8d0ac627Update callback example to use takeDelimiterInclusive for Zig 0.15.2 delimiter behavior change.Foke Singh2025-12-12 14:00:50 +0000
7e81d022ffAdd implicit dependency on hf_transfer for tools/hf.Tarry Singh2025-12-08 09:24:52 +0000
fa886db3bbDisable ZML aio torch pickle test pending implementation fix due to protocol support limitation in version 0.15.2.Tarry Singh2025-12-03 10:57:27 +0000
ba5043ddf8Switch ZML to use .host_pinned memory on ROCm, addressing missing host_unpinned memory error.Tarry Singh2025-12-01 14:34:13 +0000
7894335dc9Correct compilation errors in examples/callback.Foke Singh2025-11-26 15:23:41 +0000
b58f7ced3dFix $ORIGIN handling in runtimes/neuron by escaping $ since zigopts does not expand Make variables.Tarry Singh2025-11-24 12:04:56 +0000
993cbffd22Upgrade workspace Zig version to 0.15.2.Tarry Singh2025-11-21 12:12:36 +0000
d3baeed676Add -fllvm flag to the MNIST example BUILD configuration.Foke Singh2025-11-20 15:28:47 +0000
3d3a0ea463Enable -fllvm flag for CUDA runtime, upb, and ZML Bazel targets.Tarry Singh2025-11-11 13:02:40 +0000
94c5bbcfa6Update workspace to use rules_zig archive_override before version 0.12.2, fixing shared_library and PJRT loading and ensuring os/log.h inclusion.Tarry Singh2025-11-06 12:36:22 +0000
91f1c3b7aapjrt/cpu: fix compilation when CPU is absent by ensuring CcCompilationContext is non-empty for Zig modules accessing the c moduleTarry Singh2025-10-30 09:01:36 +0000
b67a3f9a0aworkspace: switch to libxev from Bazel Central Registry, completing migration away from custom registryTarry Singh2025-10-29 17:43:22 +0000
675cefaf26Patch http_deb_archive rule and improve stdx utilities and ZML callback handlingTarry Singh2025-10-20 10:24:27 +0000
20ca4a6648Update MNIST example BUILD file to use zigopts and align with new upstreamed rules_zig configuration.Foke Singh2025-10-14 16:57:38 +0000
bcd43314a4Migrate BUILD and MODULE files to upstreamed rules_zig (rename copts to zigopts) and adjust ZLS integration accordingly.Tarry Singh2025-10-13 15:26:42 +0000
7d7c124adaruntimes/cpu: update Darwin arm64 PJRT tarball checksum to match v13.0.0, enabling Bazel fetch on macOS arm64Tarry Singh2025-10-07 10:18:40 +0000
77cd21d2b2Add gpt-oss model support to core ZML components: formatting, utility functions, safetensors I/O, host buffer management, NN layer handling, and tensor operations.Tarry Singh2025-10-01 14:20:32 +0000
e1b7fc5781Add example implementation and Bazel build for OpenAI gpt-oss models (GptOss.zig, main.zig, and BUILD.bazel).Foke Singh2025-09-30 16:17:00 +0000
d45a667ee5Revamp gather API with named indices (and add gather_ variant), improve topK handling, and add Yarn rope embedding support across core modules (buffer, nn, pjrtx, quantization, shape, tensor, testing, tokenizer, torch).Tarry Singh2025-09-26 13:38:11 +0000
7264fff493Update MNIST and ModernBERT example scripts to use the new gather and topk APIs.Foke Singh2025-09-24 15:42:09 +0000
29bd1242baAdd platform tag to buffers for target identification and safety checks; include workaround for PJRT uninitialized memory handling.Tarry Singh2025-09-16 17:43:18 +0000
9aeb4e9cd0Update docs and example to demonstrate adding the platform tag to buffers when adding weights.Foke Singh2025-09-11 11:15:37 +0000
12efc763d5Update modernbert example to reflect removal of old writer interface APIs.Foke Singh2025-09-09 10:54:06 +0000
090d7748d5Update docs and example scripts (including test files) to remove async helpers asynk, asyncc, awaitt, and await_Foke Singh2025-09-02 10:27:40 +0000
6e15123fb3Remove obsolete async symbols (asynk, asyncc, awaitt, await_) from core, runtime, and aio modules.Tarry Singh2025-08-29 11:03:59 +0000
7913c00d70Implement buffer‑ID based loading by moving tensor ID handling into BufferStore, fix zml.call tag hashing, and expose CPU device count.Tarry Singh2025-08-28 14:39:21 +0000
6e7617918dUpdate llama and simple_layer examples to use BufferStore tensor IDs, new CPU device count API, and fix zml.call tag hashing.Foke Singh2025-08-22 17:55:03 +0000
cc969bd532Add experimental zml.callback API (renamed from custom_call) and fix tensor.print(); update PJRT bindings, host buffer utilities, and related core ZML modules.Tarry Singh2025-08-20 10:27:54 +0000
1fa056a790Add example and Bazel build for zml.callback demonstrating a manual CUDA kernel invocation.Foke Singh2025-08-13 12:14:53 +0000
01da2184fexla: bump to commit b3fbfee, temporarily disable libnvptxcompiler due to missing support in PjRT CUDA plugin v13.0, add nvshmem to sandbox for PjRT CUDA pluginTarry Singh2025-08-12 13:32:18 +0000
f35119f768ci: revamp commit workflow, deduplicate workflows, fix Linux linking issue introduced by upb, resolve duplicated _start symbol in test targets, temporarily use -c opt to bypass static archive bugTarry Singh2025-08-08 13:51:27 +0000
9e3cd6d616bump runtimes/* code to Zig 0.15.1, restore PyTorch loader using std.fs.File, update CI zig fmt, remove stdx.io, note remaining issues with Neuron and CUDA debug buildsTarry Singh2025-08-07 15:09:27 +0000
0ed7f5c907Add Zig example programs for LLaMA, ModernBERT, and SimpleLayer, including a Bazel BUILD file for the LLaMA example.Foke Singh2025-07-29 16:07:11 +0000
e1ee340306runtimes/cuda: implement zmlxcuda in ZigTarry Singh2025-07-08 09:25:25 +0000
c488b634fcruntimes/rocm: implement zmlxrocm in ZigTarry Singh2025-07-07 16:48:07 +0000
a512b9c8a2Update docs and example BUILD files to reflect the switch from zig_cc_binary to zig_binary.Foke Singh2025-07-04 15:10:46 +0000
cf00506dbbSwitch workspace build rules from zig_cc_binary to zig_binary, removing the hack and using the C linker directly.Tarry Singh2025-07-03 15:10:36 +0000
7fb02e1888Update docs: add deployment guide, Hugging Face token instructions, getting‑started tutorial, and include a Bazel lock example.Foke Singh2025-06-05 13:18:14 +0000
f9280b1069Add Bazel MODULE and tool build configuration (BUILD, requirements) for packaging the HF CLI.Tarry Singh2025-05-30 16:42:57 +0000
111afcdd95examples/llama: switch to --hf-model-path flagFoke Singh2025-05-28 13:21:00 +0000
5a49a3e8caUpdate examples MODULE.bazel.lock to reflect the added sandbox neuron dependencies.Foke Singh2025-05-22 10:10:14 +0000
1a2b862ec2Add sandbox neuron dependencies: define a trampoline PJRT, create an empty repository for distroless deps, and update Bazel build files and Zig/C sources accordingly.Tarry Singh2025-05-19 17:35:33 +0000
9e66d5a06aUpdate examples/MODULE.bazel.lock to reflect new dependencies.Foke Singh2025-05-16 11:26:13 +0000
55c5b540f8Add XLA 20250718.0‑6319f0d with ROCm 6.4.1 support, update Bazel module files and runtime configs, and apply migration, FFI‑handler and header‑cleanup patches.Tarry Singh2025-05-12 12:10:27 +0000
ed5ae31338runtimes/rocm: fetch libdrm from amdgpu repository and add amdgpu.ids layerTarry Singh2025-04-30 15:53:51 +0000
47146fc088workspace: remove glibc version constraint, update rules_rust, and re-enable frame pointers following sandboxing changes.Tarry Singh2025-04-29 12:09:13 +0000
e7323be10bruntimes/rocm: switch to in-process LLD, removing the need for sandboxed lld.Tarry Singh2025-04-23 11:43:18 +0000
02ba41ea57Update examples/MODULE.bazel to bump rules_zig version and fix static‑lib inclusion for non‑exe targets.Foke Singh2025-04-21 13:51:52 +0000
f5f3dc6b1bUpdate rules_zig to 20250714.0-b14a4f1 and fix static‑lib inclusion in runfiles for non‑exe targets.Tarry Singh2025-04-18 12:09:38 +0000
7d9fdf94e7runtimes/rocm: sandbox ROCm dependencies and ensure they load on the main thread due to TLS usage in static C++ destructors.Tarry Singh2025-04-14 16:38:15 +0000
8073e45894Update examples/MODULE.bazel.lock to reflect bumped hftokenizers dependency.Foke Singh2025-04-09 10:21:44 +0000
4294a4d08fBump hftokenizers dependency versions in Bazel and Cargo lockfiles (MODULE.bazel.lock, Cargo.toml, Cargo.lock)Tarry Singh2025-04-04 12:54:33 +0000
78d7b672e7runtimes/cpu: sandbox CPU PJRT plugin, simplifying as there are no additional NEEDED dependencies.Tarry Singh2025-04-03 11:57:46 +0000
2d321d232druntimes/cuda: sandbox CUDA dependencies by removing them from the leaf binary, sandboxing the dependency graph, marking dlopen direct dependencies as NEEDED, setting RPATH to the sandbox, loading the PJRT plugin from the sandbox, and enabling weak CUDA symbols without direct linking.Tarry Singh2025-03-26 11:18:29 +0000
a5420068b1pjrt: emit warning instead of panic when FFI Extension is missing (e.g., on TPU).Tarry Singh2025-03-24 09:40:44 +0000
dc121fce4fUpdate example MODULE.bazel and lockfile to reflect toolchains_llvm_bootstrapped bump to 0.2.4.Foke Singh2025-03-20 12:17:30 +0000
907577525fUpdate MODULE.bazel and lockfile to bump toolchains_llvm_bootstrapped to version 0.2.4.Tarry Singh2025-03-18 11:47:22 +0000
f27a524f31Update rules_zig: add zig_srcs target, fix source handling bug, clean up BUILD files, adjust async/coro.zig tests, and disable nemo and yaml model loaders.Tarry Singh2025-03-13 12:27:21 +0000
6fc1148206async/coro: make coroutines unwindable by zeroing the initial stack region, preventing random unwinding behavior and SIGSEGV during _Unwind_Backtrace.Tarry Singh2025-03-10 16:25:45 +0000
f63c673f45bazel: add RPATH manipulation to patchelfTarry Singh2025-03-05 11:56:40 +0000
9488672d4bworkspace: bump xla to version 20250710.0-22ea002Tarry Singh2025-03-04 17:12:34 +0000
fa0ed045efruntimes/cuda: downgrade cuda and cudnnTarry Singh2025-02-28 17:36:12 +0000
8456a0d073zml/pjrt: add binding for PJRT_Device_MemoryStats.Tarry Singh2025-02-19 12:14:05 +0000
a580f2a398Async: use stronger memory ordering to prevent potential segfaults due to ordering issues.Tarry Singh2025-02-18 11:38:56 +0000
4d6d975dc0Patch aio.zig: update loadBuffersWithPrefix argument type to match the conditional type of loadBuffers init_args.Tarry Singh2025-02-13 09:48:13 +0000
af8844c1f1Add model prefix support when loading a model from safetensors, enabling use of a specific model prefix (e.g., ModernBertModel) instead of the full model. Tested with the text embeddings server project.Tarry Singh2025-02-12 13:18:27 +0000
Foke Singh
Tarry Singh