Commit graph

  • 2edb45e2c5 chore: set macOS deployment target to 26.0 main Snider 2026-02-26 05:38:53 +00:00
  • 580e547848 chore: bump forge.lthn.ai dep versions to latest tags Snider 2026-02-26 05:34:22 +00:00
  • 802b7660f2 feat: update mlxlm bridge for mlx-lm 0.30.7 API v0.1.0 Snider 2026-02-23 18:37:03 +00:00
  • 71fe4bb5ac fix: add Detach/Free calls to reduce Metal GPU memory retention Snider 2026-02-23 18:36:57 +00:00
  • c1baeb9254 Merge pull request 'chore: Go 1.26 modernization' (#2) from chore/go-1.26-modernization into main Charon 2026-02-24 18:01:47 +00:00
  • eeec1fe372
    chore: use slices.Sorted(maps.Keys()) for ordered iteration Claude 2026-02-24 16:32:47 +00:00
  • 85775dc082
    chore: use range-over-integer (Go 1.22+) Claude 2026-02-24 16:32:00 +00:00
  • 694e78ca34
    chore: sort.Slice → slices.SortFunc Claude 2026-02-24 16:29:48 +00:00
  • 9f6dd9d4eb
    chore: fmt.Errorf(static) → errors.New Claude 2026-02-24 16:28:27 +00:00
  • 421d0c42ff
    docs: archive completed backend-abstraction and batch-inference plans Claude 2026-02-24 13:51:21 +00:00
  • c0f07478c8 docs: document InspectAttention KV cache extraction in architecture guide Snider 2026-02-23 12:34:31 +00:00
  • f7553eb277 chore: bump go-inference to v0.0.3 v0.0.3 Snider 2026-02-23 06:59:59 +00:00
  • 30eee2525f chore: add Go repo norms (badges, contributing, lint, taskfile, editorconfig) Snider 2026-02-23 06:45:42 +00:00
  • 51ac442a09 fix: add deterministic GPU memory cleanup across inference paths v0.0.2 Snider 2026-02-23 04:37:51 +00:00
  • 208f76b067
    ci: add Forgejo Actions test and security scan workflows Claude 2026-02-23 03:28:08 +00:00
  • c2177f754a feat: implement AttentionInspector via KV cache extraction after prefill Snider 2026-02-23 00:25:09 +00:00
  • 9a48774538
    chore: sync workspace dependency versions Claude 2026-02-22 21:41:04 +00:00
  • 5004ac258a refactor: apply go fix modernizers for Go 1.26 v0.0.1 Snider 2026-02-22 21:00:16 +00:00
  • fc27c2cd27 chore: go mod tidy for 1.26.0 Snider 2026-02-22 20:35:59 +00:00
  • 60f9a5b740 chore: bump go directive to 1.26.0 Snider 2026-02-22 20:33:48 +00:00
  • 2f0be488dd chore: pin forge deps to v0.0.1 tags for Go 1.26 compat Snider 2026-02-22 20:15:06 +00:00
  • 094d82f535 chore: use workspace-resolved versions, drop replace directives Snider 2026-02-22 19:49:36 +00:00
  • 2292557fd6 chore: vendor MLX C headers for Go module consumers Snider 2026-02-21 19:14:04 +00:00
  • 1c2f13fd9d chore: resolve go-inference from forge, remove local replace Snider 2026-02-21 19:08:51 +00:00
  • 98acc11e23 docs: add README with quick start and docs links Snider 2026-02-20 15:11:20 +00:00
  • 1ea90b03b4 docs: graduate TODO/FINDINGS into production documentation Snider 2026-02-20 15:03:39 +00:00
  • 015ebbc2ac chore: add gitignore entries for IDE/build artifacts Snider 2026-02-20 09:15:25 +00:00
  • 757a241f59 feat(mlxlm): Phase 5.5 — subprocess backend using Python mlx-lm Snider 2026-02-20 09:02:30 +00:00
  • 887c221974 docs: write detailed mlxlm subprocess backend spec (Phase 5.5) Snider 2026-02-20 08:58:19 +00:00
  • d1fb26d51e docs: expand package doc with workflow examples Snider 2026-02-19 23:44:07 +00:00
  • d7c8f176f0 docs: mark embed-friendly model loading complete in TODO Snider 2026-02-19 23:37:58 +00:00
  • dd49b4afb6 feat: add model discovery test and update TODO Snider 2026-02-19 23:37:37 +00:00
  • ceb966b66b feat(metal): expose model metadata via Info() Snider 2026-02-19 23:36:23 +00:00
  • a44e9f5789 feat(metal): add inference metrics (timing, throughput, memory) Snider 2026-02-19 23:34:40 +00:00
  • 197326bd22 docs: mark batch inference API complete in TODO.md Snider 2026-02-19 23:28:40 +00:00
  • 5644857034 feat(metal): implement batch inference (Classify, BatchGenerate) Snider 2026-02-19 23:28:15 +00:00
  • ce1acef462 docs: batch inference API design (Phase 5) Snider 2026-02-19 23:18:38 +00:00
  • e3fbc221ce feat(metal): add mixed precision training via LoRAConfig.DType (Phase 3) Snider 2026-02-19 23:13:49 +00:00
  • fa08ed1e2a test(metal): validate gradient checkpointing with real model (Phase 3) Snider 2026-02-19 23:11:15 +00:00
  • fb0692baf3 test(metal): add LoRA end-to-end training pipeline test (Phase 3) Snider 2026-02-19 23:09:16 +00:00
  • 19c4823b04 feat(metal): add Llama 3 model support (Llama 3.1 8B validated) Snider 2026-02-19 23:06:43 +00:00
  • 535b04d5d6 feat(metal): add Qwen2 model support (DeepSeek R1 validated) Snider 2026-02-19 21:55:56 +00:00
  • a2493e0242 test(metal): add model loading robustness tests (Phase 2) Snider 2026-02-19 21:49:07 +00:00
  • 18e8dca9f8 feat(metal): validate Gemma3-1B inference end-to-end (Phase 2) Snider 2026-02-19 21:44:28 +00:00
  • 443347a2f8 fix(metal): address 4 minor code review items Snider 2026-02-19 21:36:40 +00:00
  • fb95cde30c fix(metal): address 5 important code review items Snider 2026-02-19 21:31:45 +00:00
  • c96f9bd006 fix(metal): address 3 critical code review items Snider 2026-02-19 21:24:10 +00:00
  • bd5668967c docs: Virgil code review feedback after Phase 4 completion Snider 2026-02-19 21:17:43 +00:00
  • f13a8c9289 feat(metal): deterministic Close() and Array.Iter() Snider 2026-02-19 21:09:39 +00:00
  • f2ca7fe188 docs(cpp): add on-demand research tasks for CLion Claude Snider 2026-02-19 21:01:45 +00:00
  • 754d6e2f93 fix(metal): error handling audit — propagate MLX errors instead of swallowing Snider 2026-02-19 20:59:37 +00:00
  • ff01175a62 bench(metal): add 29 benchmarks baselined on M3 Ultra Snider 2026-02-19 20:47:25 +00:00
  • 7435648f66 docs: update findings and TODO with CLion Claude research results Snider 2026-02-19 20:40:35 +00:00
  • f507620e85 docs(cpp): complete mlx-c API research — all tasks done Snider 2026-02-19 20:40:00 +00:00
  • ca6b16eaf2 feat(metal): bind memory diagnostics and device info Snider 2026-02-19 20:39:51 +00:00
  • f39126f6bd feat(metal): bind CumSum, implement TopP and MinP sampling Snider 2026-02-19 20:39:44 +00:00
  • df0b300b1a fix(metal): auto-contiguous data access for non-contiguous arrays Snider 2026-02-19 20:39:36 +00:00
  • bff97ccf19 feat(api): migrate to go-inference shared interfaces Snider 2026-02-19 20:15:42 +00:00
  • 95d92fffff docs: update project docs for backend abstraction completion Snider 2026-02-19 20:07:01 +00:00
  • eb8dee31bf test(api): integration tests for public LoadModel + Generate Snider 2026-02-19 20:05:14 +00:00
  • 4d1bff3d78 refactor(api): clean root package — interfaces only, metal auto-registered Snider 2026-02-19 20:04:19 +00:00
  • c612c3e060 refactor(metal): move all tests to internal/metal (148 tests passing) Snider 2026-02-19 20:00:02 +00:00
  • 08976aa504 refactor(metal): flatten model, tokenizer, sample, cache into internal/metal Snider 2026-02-19 19:51:14 +00:00
  • a669d1d9c1 refactor(metal): move nn, io, grad, lora, optim to internal/metal Snider 2026-02-19 19:39:58 +00:00
  • d6a49544bd refactor(metal): move ops, slice, random, fast, compile to internal/metal Snider 2026-02-19 19:39:49 +00:00
  • 1cf5178c80 refactor(metal): move dtype, array, metal, stream to internal/metal Snider 2026-02-19 19:34:38 +00:00
  • 643df757e0 feat(api): define public TextModel, Backend, and options interfaces Snider 2026-02-19 19:32:00 +00:00
  • f7a57657d4 chore: gitignore dist/ (CMake install output) Snider 2026-02-19 19:30:23 +00:00
  • 97d9041455 docs(plan): fold Virgil review into design and implementation plan Snider 2026-02-19 19:25:05 +00:00
  • 28e2a07316 docs(plan): backend abstraction implementation plan (12 tasks) Snider 2026-02-19 19:14:59 +00:00
  • c881813872 docs(design): backend abstraction with internal/metal reorganisation Snider 2026-02-19 19:12:04 +00:00
  • 3b155f6ca7 chore(cpp): set up standalone CLion project for parallel C++ sessions Snider 2026-02-19 18:49:16 +00:00
  • 28b5c5bb47 test(sub-packages): add 33 tests for cache, sample, tokenizer Snider 2026-02-19 18:40:49 +00:00
  • 37abc496ba test(core): add 86 tests for ops, array, nn, fast kernels Snider 2026-02-19 18:37:30 +00:00
  • 40cbdd7201 docs(todo): add ecosystem integration and Go 1.26 wishlist items Snider 2026-02-19 18:18:32 +00:00
  • cae7ef05e8 feat: extract go-mlx from go-ai as standalone Metal inference package Snider 2026-02-19 17:57:37 +00:00