| COMMIT |
1.00 |
runner: fix swallowed error in allocModel graph reservation |
|
Commit message contains explicit AI assi |
2026-03-14 |
| COMMIT |
0.90 |
doc: update vscode doc (#15064) |
|
Contains AI collaboration trailer (Co-au |
2026-03-26 |
| PR |
0.90 |
fix: pin 10 unpinned action(s),extract 1 unsafe expression(s |
|
Very formal, generic introduction, typic |
2026-03-26 |
| PR |
0.80 |
fix: security assessment identifies authentication m... in a |
|
Has AI security scanner and automated fi |
2026-03-30 |
| PR |
0.30 |
fix: memory leak when when vocab loading from file failed |
|
Uses template headings with perfect gram |
2026-03-30 |
| PR |
0.25 |
docs: add kbot integration |
|
Lists, product links, and domain stats; |
2026-04-01 |
| PR |
0.20 |
docs: clarify num_ctx parameter description |
|
Clear, technical, and contextualized; hu |
2025-11-20 |
| PR |
0.20 |
cuda: graceful OOM fallback when creating events during part |
|
In-depth hardware context and bug repro; |
2026-03-04 |
| PR |
0.20 |
create: fix parser/renderer mapping for qwen3 variants |
|
Detailed explanation, contextual and tec |
2026-04-02 |
| PR |
0.20 |
launch: replace deprecated OPENAI_BASE_URL with config.toml |
|
Detailed technical content, uses domain- |
2026-03-24 |
| PR |
0.20 |
fix(registry): avoid panic in statusCodeRecorder on non-Flus |
|
Somewhat formal but mixes technical summ |
2026-04-01 |
| PR |
0.20 |
Draft: api: support multimodal embedding |
|
Direct status update, mentions links, ty |
2026-03-31 |
| PR |
0.20 |
Preserve thought signatures in tool calls and messages (fix |
|
Technical explanation, some structure, b |
2026-03-06 |
| PR |
0.20 |
fix(app): prevent sidebar animation on initial load |
|
Detailed technical summary, no AI-like p |
2026-03-31 |
| PR |
0.20 |
Windows installer options |
|
Free-text is detailed but natural, sligh |
2026-03-28 |
| PR |
0.15 |
Fix: Resolve incorrect GGUF magic validation in fs/gguf/gguf |
|
Marks sections, but changes are describe |
2026-04-01 |
| COMMIT |
0.10 |
server: preserve raw manifest bytes during pull (#15104) |
|
Detailed but technical and focused, like |
2026-03-27 |
| COMMIT |
0.10 |
tui: update chat title (#15082) |
|
Terse human-like commit message with typ |
2026-03-27 |
| COMMIT |
0.10 |
launch/vscode: prefer known vs code paths over `code` on PAT |
|
Standard technical description: terse an |
2026-03-26 |
| COMMIT |
0.10 |
launch: hide cline integration (#15080) |
|
Very brief and informal human commit-sty |
2026-03-26 |
| COMMIT |
0.10 |
launch: hide vs code (#15076) |
|
Simple, terse, and human-like title with |
2026-03-26 |
| PR |
0.10 |
Reduce per-call tokenizer overhead by 3-5× |
|
Technical changelog, performance data; h |
2026-02-07 |
| PR |
0.10 |
app/ui: enable thinking toggle for all thinking-capable mode |
|
Technical and code-specific description; |
2026-04-02 |
| PR |
0.10 |
fix(mlx): prevent Metal GPU watchdog crash on large models |
|
Has technical specifics and context; nat |
2026-04-02 |
| PR |
0.10 |
model: normalize Unicode hyphens in model names |
|
Pain point described with nuance and ane |
2026-03-26 |
| PR |
0.10 |
docs: restore num_gpu parameter in Modelfile documentation |
|
Restoration rationale and documentation |
2026-02-04 |
| PR |
0.10 |
launch/opencode: detect `curl` installed opencode at `~/.ope |
|
Technical language, concise, uses domain |
2026-04-01 |
| PR |
0.10 |
feat: Add Support for Qwen3-vl and Qwen2.5-vl Video Mode |
|
Concise, technical, with mild formality; |
2025-11-05 |
| PR |
0.10 |
create: Clean up experimental paths, fix create from existi |
|
Brief, has typos and light domain terms, |
2026-03-06 |
| PR |
0.10 |
tokenizer: add SentencePiece-style BPE support |
|
Concise, domain-specific, minor formal p |
2026-03-31 |
| PR |
0.10 |
mlx: respect tokenizer add_bos_token setting in pipeline |
|
Technical, terse summary with inline dif |
2026-03-31 |
| PR |
0.10 |
Add missing hipBLASLt kernels and some gfx targets cleanup |
|
Uses domain jargon, concise changelog st |
2026-03-20 |
| PR |
0.10 |
fix(fileutil): use user-scoped temp directory for backups |
|
Technical breakdown, domain details, nat |
2026-03-31 |
| PR |
0.10 |
fix(discover): warn when GPU backend runner fails to start |
|
Direct, technical, uses bugfix language |
2026-03-31 |
| PR |
0.10 |
mlx: enhance multimodal pipeline with snapshot scheduling an |
|
Detailed, technical list—no AI-specific |
2026-03-31 |
| PR |
0.10 |
discover: retry CUDA probe without visibility filter when CU |
|
Technical, terse, references domain spec |
2026-03-30 |
| PR |
0.10 |
Added : OpenAPI spec: ChatStreamEvent missing usage fields i |
|
Mostly technical, but some slightly form |
2026-03-08 |
| PR |
0.10 |
feat: add no‑op think tool for structured reasoning |
|
Technical, direct, minor hedging but ove |
2026-03-29 |
| PR |
0.10 |
feat: add experimental WebGPU backend stub documentation |
|
Informal, collaborative; no ChatGPT phra |
2026-03-29 |
| PR |
0.10 |
Expose numeric parameter_count in model details |
|
Technical, issue-referenced, no AI styli |
2026-03-28 |
| PR |
0.10 |
ui: add image generation support for image-capable models |
|
Technical summary, bullet points, no AI- |
2026-03-27 |
| PR |
0.10 |
re-add OLLAMA_MAX_VRAM |
|
Informal, includes personal opinion and |
2026-02-21 |
| COMMIT |
0.08 |
api/show: overwrite basename for copilot chat (#15062) |
|
Technical, informal, domain-specific; hu |
2026-03-25 |
| COMMIT |
0.05 |
bench: add prompt calibration, context size flag, and NumCtx |
|
Technical detail, terse, domain-specific |
2026-04-02 |
| COMMIT |
0.05 |
tokenizer: add byte fallback for SentencePiece BPE encoding |
|
Technical, specific, informal commit mes |
2026-04-02 |
| COMMIT |
0.05 |
Add support for gemma4 (#15214) |
|
Mixture of technical summary and test de |
2026-04-02 |
| COMMIT |
0.05 |
anthropic: fix empty inputs in content blocks (#15105) |
|
Brief, uses domain-appropriate Go termin |
2026-03-27 |
| COMMIT |
0.05 |
launch: skip context length warning for MLX models and show |
|
Concise, omits extraneous detail, classi |
2026-03-27 |
| COMMIT |
0.05 |
app: fix false "out of date" model warnings (#15101) |
|
Contains in-depth technical details, inf |
2026-03-27 |
| COMMIT |
0.05 |
parsers: qwen3.5 streaming tool-call parsing and add regress |
|
Short, uses specific jargon, informal; c |
2026-03-27 |
| COMMIT |
0.05 |
model/parsers: close think block if tool block starts in Qwe |
|
Terse commit, focused, no AI signs. |
2026-03-27 |
| COMMIT |
0.05 |
cmd/launch: skip --install-daemon when systemd is unavailabl |
|
Commit message and description are conci |
2026-03-16 |
| COMMIT |
0.05 |
bench: improve benchmarking tool (#14240) |
|
Technical jargon and bullet point change |
2026-03-15 |
| COMMIT |
0.05 |
config: use native OpenClaw Ollama onboarding (#14829) |
|
Brief, domain-focused text with technica |
2026-03-13 |
| COMMIT |
0.05 |
server: decompress zstd request bodies in cloud passthrough |
|
Technical language and error context; no |
2026-03-13 |
| COMMIT |
0.05 |
middleware: handle non-JSON error responses gracefully (#148 |
|
Brief technical explanation with domain |
2026-03-13 |
| COMMIT |
0.05 |
anthropic: close thinking block before tool_use when no text |
|
Detailed root cause and fix; clear domai |
2026-03-13 |
| COMMIT |
0.05 |
cmd: still populate `ollama ls` when using `ollama run |
|
Short, technical note; lacks generic AI |
2026-03-13 |
| PR |
0.05 |
tokenizer: merge pre-tok REs to prevent multiple matches |
|
Technical RE explanation, test output, i |
2026-02-12 |
| PR |
0.05 |
api: add audios field for audio input in multimodal models |
|
Detailed, changelog-style human writing |
2026-04-02 |
| PR |
0.05 |
scripts: fix installer exiting early when nvidia-smi is foun |
|
Technical explanation with natural tone, |
2026-04-02 |
| PR |
0.05 |
create: add vision capability detection and fix capability i |
|
Technical, brief, and specific with norm |
2026-03-17 |
| PR |
0.05 |
integration: improve vision test robustness and add thinking |
|
Technical, concise, and contains natural |
2026-03-31 |
| PR |
0.05 |
cmd: improve multi-select sorting and selection status |
|
Structured summary but informal and cont |
2026-04-01 |
| PR |
0.05 |
Add conversation context usage indicator |
|
Technical, clear, and with developer det |
2026-03-27 |
| PR |
0.05 |
tools: warn when model tool call names don't match registere |
|
Technical, informal, and contains specif |
2026-03-21 |
| PR |
0.05 |
server: add download bandwidth rate limiting via OLLAMA_MAX_ |
|
Clear, domain-specific, terse; no AI hal |
2026-04-02 |
| PR |
0.05 |
server: check available disk space before model download |
|
Technical with natural imperfections; no |
2026-04-02 |
| PR |
0.05 |
logutil: support JSON log output via OLLAMA_LOG_FORMAT |
|
Succinct, technical, and pragmatically w |
2026-04-02 |
| PR |
0.05 |
cmd: show freed disk space after ollama rm |
|
Template format but free text is direct |
2026-04-02 |
| PR |
0.05 |
llm: add automatic MOE expert weight offloading |
|
Uses technical terms and natural enginee |
2026-04-02 |
| PR |
0.05 |
Add reranking support |
|
Informal, detailed, references project s |
2026-02-09 |
| PR |
0.05 |
sample: add segment-level repetition loop detection |
|
Technical, uses abbreviations, and prese |
2026-04-02 |
| PR |
0.05 |
docs/integrations: add haiku.rag |
|
Brief, project-specific, informal tone. |
2026-04-02 |
| PR |
0.05 |
docs: add community integration to README |
|
Informal, personalized tone and emotive |
2026-03-24 |
| PR |
0.05 |
Add token usage for cloud model streaming (inject stream_opt |
|
Terse, domain-specific, manually written |
2026-04-02 |
| PR |
0.05 |
launch: improve multi-select for already added models |
|
Terse, informal, uses 'ux' and minimal e |
2026-03-28 |
| PR |
0.05 |
create: move safetensor model creation server-side |
|
Technical summary, domain details; lacks |
2026-03-19 |
| PR |
0.05 |
server: preserve raw manifest bytes during pull |
|
Technical, detail-oriented, not overly f |
2026-03-27 |
| PR |
0.05 |
anthropic: fix empty inputs in content blocks |
|
Same as #1; domain-specific, brief, info |
2026-03-27 |
| COMMIT |
0.00 |
app: use the same client for inference and other requests (# |
|
Concise, plain commit message; no AI tex |
2026-04-02 |
| COMMIT |
0.00 |
app: add launch page for a simple way to launch integrations |
|
Very brief, no stylistic evidence of AI |
2026-04-02 |
| COMMIT |
0.00 |
launch: replace deprecated OPENAI_BASE_URL with config.toml |
|
Brief, domain-specific commit message; n |
2026-04-01 |
| COMMIT |
0.00 |
tokenizer: add SentencePiece-style BPE support (#15162) |
|
Contains review comments and technical j |
2026-04-01 |
| COMMIT |
0.00 |
mlx: respect tokenizer add_bos_token setting in pipeline (#1 |
|
Thorough technical changelog, specific j |
2026-03-31 |
| COMMIT |
0.00 |
docs: update pi docs (#15152) |
|
Typical doc update summary; too brief fo |
2026-03-31 |
| COMMIT |
0.00 |
cmd: set OpenCode default model in config (#15127) |
|
Concise commit message, entirely human s |
2026-03-29 |
| COMMIT |
0.00 |
model: add qwen3-next compatibility for legacy ssm_in projec |
|
Brief, domain-specific commit; no AI sig |
2026-03-29 |
| COMMIT |
0.00 |
revert context length warnings change (#15121) |
|
Terse, typical revert commit phrasing. |
2026-03-28 |
| COMMIT |
0.00 |
launch: improve multi-select for already added models (#1511 |
|
Standard commit summary, no AI signals. |
2026-03-28 |
| COMMIT |
0.00 |
launch: auto-install pi and manage web-search lifecycle (#15 |
|
Succinct and technical, no AI phrasing d |
2026-03-28 |
| COMMIT |
0.00 |
mlx: fix vision capability + min version (#15106) |
|
Typical terse commit style, no AI indica |
2026-03-28 |
| COMMIT |
0.00 |
anthropic: fix KV cache reuse degraded by tool call argument |
|
Technical explanation, domain jargon, no |
2026-03-09 |
| COMMIT |
0.00 |
ci: harden cuda include path handling (#15093) |
|
Terse, technical commit message with nat |
2026-03-27 |
| COMMIT |
0.00 |
launch: warn when server context length is below 64k for loc |
|
Minimal, informal explanation; human co- |
2026-03-27 |
| COMMIT |
0.00 |
ci: include mlx jit headers on linux (#15083) |
|
Concise, domain-specific phrasing; no AI |
2026-03-27 |
| COMMIT |
0.00 |
mlxrunner: combine setStateRaw and setStateDetached into set |
|
Concise, technical commit message withou |
2026-03-25 |
| COMMIT |
0.00 |
mlxrunner: schedule periodic snapshots during prefill |
|
Detailed but natural human technical des |
2026-03-24 |
| COMMIT |
0.00 |
mlxrunner: improve eviction and LRU tracking |
|
Detailed, technical explanation with cle |
2026-03-24 |
| COMMIT |
0.00 |
mlx: fix KV cache snapshot memory leak |
|
Technical detail and domain jargon in co |
2026-03-25 |
| COMMIT |
0.00 |
ggml: force flash attention off for grok |
|
Brief, terse CLI change with target doma |
2026-03-25 |
| COMMIT |
0.00 |
cmd: ollama launch vscode (#15060) |
|
Has Co-authored-by, terse commit; clearl |
2026-03-25 |
| COMMIT |
0.00 |
tui: add left arrow back navigation in model selector (#1494 |
|
Short, informal, technical phrasing; hum |
2026-03-25 |
| COMMIT |
0.00 |
ci: fix windows cgo compiler error (#15046) |
|
Commit is terse and domain-specific with |
2026-03-24 |
| COMMIT |
0.00 |
integration: improve ability to test individual models (#149 |
|
— |
2026-03-24 |
| COMMIT |
0.00 |
mlx: add mxfp4/mxfp8/nvfp4 importing (#15015) |
|
— |
2026-03-24 |
| COMMIT |
0.00 |
mlxrunner: panic on double unpin |
|
Very terse, domain-specific commit messa |
2026-03-20 |
| COMMIT |
0.00 |
mlxrunner: show time since last used in cache dump tree |
|
Concise, human-written technical change. |
2026-03-19 |
| COMMIT |
0.00 |
mlxrunner: support partial match on pure transformer caches |
|
Technical explanation using domain jargo |
2026-03-19 |
| COMMIT |
0.00 |
docs: update Claude Code with Telegram guide (#15026) |
|
Standard release note; brief and human-t |
2026-03-23 |
| COMMIT |
0.00 |
mlx: update as of 3/23 (#14789) |
|
Changelog has casual tone, uses domain v |
2026-03-23 |
| COMMIT |
0.00 |
docs: nemoclaw integration (#14962) |
|
Commit has domain abbreviations, informa |
2026-03-20 |
| COMMIT |
0.00 |
launch: skip openclaw gateway health check when no daemon in |
|
Commit is terse, domain-specific, lacks |
2026-03-20 |
| COMMIT |
0.00 |
llm, mlxrunner: fix done channel value consumed by first rec |
|
Technical commit, concise language, doma |
2026-03-18 |
| COMMIT |
0.00 |
mlx: fix subprocess log deadlock |
|
Technical explanation, domain-specific d |
2026-03-18 |
| COMMIT |
0.00 |
add ability to turn on debug request logging (#14106) |
|
Debugging-focused, brief, domain-specifi |
2026-03-20 |
| COMMIT |
0.00 |
parsers: robust xml tool repair (#14961) |
|
Direct language, co-author trailer, tech |
2026-03-19 |
| COMMIT |
0.00 |
app: fix desktop app stuck loading when OLLAMA_HOST is an un |
|
Template style, minimal description, no |
2026-03-19 |
| COMMIT |
0.00 |
mlxrunner: share KV cache across conversations with common p |
|
Technical jargon, detailed and informal, |
2026-03-05 |
| COMMIT |
0.00 |
mlxrunner: fix Slice(0, 0) returning full dimension instead |
|
Technical details, informal tone, domain |
2026-03-18 |
| COMMIT |
0.00 |
cmd/launch: skip redundant config writes when model unchange |
|
Brief, domain-specific, concise commit s |
2026-03-18 |
| COMMIT |
0.00 |
docs: update claude code and openclaw for web search (#14922 |
|
Template style, terse, technical referen |
2026-03-18 |
| COMMIT |
0.00 |
launch: fix openclaw not picking up newly selected model (#1 |
|
Technical context, concise explanation, |
2026-03-18 |
| COMMIT |
0.00 |
docs: update minimax-m2.5 references to m2.7 (#14942) |
|
Standard commit message, concise and tec |
2026-03-18 |
| COMMIT |
0.00 |
tui: fix signin on headless Linux systems (#14627) |
|
Uses domain-specific language and inform |
2026-03-18 |
| COMMIT |
0.00 |
launch: disable claude attribution header to preserve KV cac |
|
Human tone; specific, clear technical ex |
2026-03-17 |
| COMMIT |
0.00 |
cmd/launch: check for both npm and git before installing Ope |
|
Brief, domain-specific, no signs of AI t |
2026-03-18 |
| COMMIT |
0.00 |
launch: register websearch for openclaw (#14914) |
|
Minimal content, lacks AI phrasing or to |
2026-03-17 |
| COMMIT |
0.00 |
cloud_proxy: for the web_search legacy path, flush on newlin |
|
Casual tone, technical explanation; clea |
2026-03-17 |
| COMMIT |
0.00 |
mlx: quantized embeddings, fast SwiGLU, and runtime fixes (# |
|
Commit is terse and uses domain jargon; |
2026-03-17 |
| COMMIT |
0.00 |
mlx: add prequantized tensor packing + changes for qwen35 (# |
|
Technical, jargon-rich, informal; no AI |
2026-03-17 |
| COMMIT |
0.00 |
sched: Model eviction for MLX |
|
Informal, specific, technical detail; la |
2026-03-02 |
| COMMIT |
0.00 |
launch: fix web search, add web fetch, and enable both for l |
|
Short, technical, and informal; typical |
2026-03-16 |
| COMMIT |
0.00 |
cmd/launch: add guards for headless mode (#14837) |
|
Extremely terse commit message, no AI si |
2026-03-14 |
| COMMIT |
0.00 |
rocm: doc driver constraints (#14833) |
|
Very terse, doc update; too minimal for |
2026-03-13 |
| COMMIT |
0.00 |
launch: remove warning for default policy (#14830) |
|
Short and terse; typical human commit me |
2026-03-13 |
| COMMIT |
0.00 |
launch: add compact window for claude code (#14823) |
|
Terse commit message, domain-specific co |
2026-03-13 |
| COMMIT |
0.00 |
api/create: always propagate `:cloud` source for cloud model |
|
Includes technical detail and informal e |
2026-03-13 |
| COMMIT |
0.00 |
launch: fix headless --yes integration flow and policy scopi |
|
Short, technical, and informal phrasing |
2026-03-13 |
| COMMIT |
0.00 |
docs: document reasoning_effort support in OpenAI-compatible |
|
Uses domain-specific language and concis |
2026-03-13 |
| COMMIT |
0.00 |
cmd: TUI model ordering (#14814) |
|
Very terse, domain-specific commit |
2026-03-13 |
| COMMIT |
0.00 |
server: remove experimental aliases support (#14810) |
|
Short technical phrase, no AI markers |
2026-03-13 |
| COMMIT |
0.00 |
cmd: refactor tui and launch (#14609) |
|
Brief, domain-specific with technical ja |
2026-03-13 |
| COMMIT |
0.00 |
ci: fix missing windows zip file (#14807) |
|
Contains technical explanation and infor |
2026-03-12 |
| COMMIT |
0.00 |
mlx: perf improvements (#14768) |
|
Technical, uses abbreviations and inform |
2026-03-12 |
| COMMIT |
0.00 |
middleware: enable local tool model for web search (#14787) |
|
Brief technical phrasing, no AI indicato |
2026-03-11 |
| COMMIT |
0.00 |
openai: split mixed thinking stream chunks via ToChunks (#14 |
|
Commit messages are concise and domain-s |
2026-03-11 |
| COMMIT |
0.00 |
mlx: only log load errors when MLX is needed (#14764) |
|
Technical reason for logging change; con |
2026-03-11 |
| COMMIT |
0.00 |
MLX: harden for init failures (#14777) |
|
Uses informal tone and domain details; n |
2026-03-11 |
| COMMIT |
0.00 |
Revert "runner: add token history sampling parameters to oll |
|
Standard revert with precise reference; |
2026-03-11 |
| COMMIT |
0.00 |
config: qwen3.5 recommendations (#14758) |
|
Brief, domain-specific wording; typical |
2026-03-11 |
| COMMIT |
0.00 |
cloud_proxy: send ollama client version (#14769) |
|
Specific technical history; naturally ph |
2026-03-10 |
| COMMIT |
0.00 |
docs: update claude code docs (#14770) |
|
Brief doc update, matches human commit s |
2026-03-10 |
| COMMIT |
0.00 |
server: add experimental web search and web fetch routes (#1 |
|
Short, domain term 'experimental'; human |
2026-03-10 |
| COMMIT |
0.00 |
ci: Fix windows build (#14754) |
|
Technical explanation; informal, Go-spec |
2026-03-10 |
| COMMIT |
0.00 |
MLX: add header vendoring and remove go build tag (#14642) |
|
Detailed changelog with domain jargon; h |
2026-03-10 |
| COMMIT |
0.00 |
mlx: get parameters from modelfile during model creation (#1 |
|
Brief commit msg, technical, no AI trait |
2026-03-09 |
| PR |
0.00 |
mlx: add op wrappers for Conv2d, Pad, activations, trig, and |
|
Concise summary, list structure, human c |
2026-03-17 |
| PR |
0.00 |
Gemma4 on MLX |
|
WIP note, carries PR refs; too terse for |
2026-04-02 |
| PR |
0.00 |
test: integration test hardening |
|
Terse, informal, and domain-specific wor |
2025-12-19 |
| PR |
0.00 |
bench: add prompt calibration, context size flag, and NumCtx |
|
Technical, dense, and domain-specific wi |
2026-03-31 |
| PR |
0.00 |
linux: enhance docker build for mlx arm64 |
|
Short, informal, developer-focused langu |
2026-03-31 |
| PR |
0.00 |
tokenizer: add byte fallback for SentencePiece BPE encoding |
|
Domain fix with concise, technical wordi |
2026-04-02 |
| PR |
0.00 |
Add support for gemma4 |
|
Extremely terse and typical of human com |
2026-04-02 |
| PR |
0.00 |
app: use the same client for inference and other requests |
|
Terse, domain-specific, informal; classi |
2026-04-02 |
| PR |
0.00 |
app: add launch tab for a simple way to launch integrations |
|
Casual tone, specific context, no ChatGP |
2026-03-31 |
| PR |
0.00 |
Add model eval metrics to /metrics |
|
Code and metric listings; zero AI stylis |
2025-06-22 |
| PR |
0.00 |
api: preserve thought_signature in ToolCall for Gemini 3 too |
|
Concise and technical, using domain-spec |
2026-03-03 |
| PR |
0.00 |
convert: support new Gemma4 audio_tower tensor naming |
|
Extremely terse commit; standard human e |
2026-04-02 |
| PR |
0.00 |
Feature/bonsai q1 0 support |
|
Extremely terse; possibly template/place |
2026-04-02 |
| PR |
0.00 |
kvcache: add TurboQuant rotation-enhanced KV cache compressi |
|
Technical summary, concise, uses domain |
2026-03-27 |
| PR |
0.00 |
Draft: ml: support openvino |
|
Technical and log-focused, with casual i |
2026-04-02 |
| PR |
0.00 |
Claude/pending feature details 5 id7b |
|
Vague, but not AI-like; could be human p |
2026-04-01 |
| PR |
0.00 |
Fix: verify actual NVIDIA GPU presence before exiting instal |
|
Direct, detailed bugfix description with |
2026-04-01 |
| PR |
0.00 |
mlxrunner: tokenize prompts in request handler goroutines |
|
Direct technical description, informal s |
2026-04-01 |
| PR |
0.00 |
README: add iOS chat client |
|
Brief and neutral README update; lacks A |
2026-03-17 |
| PR |
0.00 |
Upgrades of multiple libraries to fix security issues |
|
Plain, very terse style, no extraneous p |
2025-11-28 |
| PR |
0.00 |
docs: update pi docs |
|
Extremely terse, classic for human docs |
2026-03-30 |
| PR |
0.00 |
pull/push: refine safetensors |
|
Detailed, technical, and informal with h |
2026-03-18 |
| PR |
0.00 |
Claude/huhul multi agent os 01 a8k2 zvf5uvnqjw85 blsn7 m |
|
No text to evaluate; only a non-descript |
2025-12-21 |
| PR |
0.00 |
mlx: fix windows dependency path resolution |
|
Generic, terse commit message; no AI sig |
2026-03-30 |
| PR |
0.00 |
ci: speed up release builds |
|
Informal tone, technical shorthand, no A |
2026-03-25 |
| PR |
0.00 |
Use LXCFS' formula to compute free memory in a cgroup |
|
Personal tone with informal trust commen |
2026-01-19 |
| PR |
0.00 |
Fix: image path not recognized correctly with escaped spaces |
|
Technical, problem-focused, clear domain |
2026-03-30 |
| PR |
0.00 |
Fix single directory selector |
|
Terse, technical, includes reviewer feed |
2025-11-12 |
| PR |
0.00 |
Fix streaming timeout during tool call composition |
|
Domain jargon, concise with casual human |
2026-03-18 |
| PR |
0.00 |
cmd: set OpenCode default model in config |
|
Minimal template content, no signs of AI |
2026-03-29 |
| PR |
0.00 |
Improve https://github.com/ollama/ollama/pull/14864 (Update |
|
Bulleted, typo in 'sugested', human-auth |
2026-03-28 |
| PR |
0.00 |
model: add qwen3-next compatibility for legacy ssm_in projec |
|
Standard reference/fix commit, clear hum |
2026-03-29 |
| PR |
0.00 |
bench: add support for OpenAI API benchmarks |
|
Tooling instructions, code blocks, domai |
2026-03-27 |
| PR |
0.00 |
launch: remove banner and warn only when backup-relevant con |
|
Concise, domain-specific changelog with |
2026-03-29 |
| PR |
0.00 |
launch: backup configs for integrations automatically |
|
Brief, informal update; no AI hints pres |
2026-03-28 |
| PR |
0.00 |
fix: prevent head-of-line blocking in scheduler during model |
|
Technical discussion with domain terms a |
2026-03-29 |
| PR |
0.00 |
kvcache: wire turboquant cache into new engine path |
|
Terse follow-up referencing previous PR, |
2026-03-29 |
| PR |
0.00 |
revert: launch: context length warnings b00bd1d 366625a8 |
|
Casual, minimal explanation; not AI-like |
2026-03-28 |
| PR |
0.00 |
runner: Replace llama CGO runner with llama-server |
|
Direct style, domain knowledge, and conc |
2026-03-28 |
| PR |
0.00 |
mac: prevent generate on cross-compiles |
|
Technical explanation with logs, clearly |
2026-03-28 |
| PR |
0.00 |
launch: auto-install pi and manage web-search lifecycle |
|
Brief, technical, bullet list; purely ta |
2026-03-28 |
| PR |
0.00 |
go.mod: resort |
|
Commit is a terse, standard practice wit |
2026-03-28 |
| PR |
0.00 |
mlx: fix vision capability + min version |
|
Very terse and uses domain abbreviation, |
2026-03-27 |
