repo stringclasses 15
values | fix_commit stringlengths 40 40 | buggy_commit stringlengths 40 40 | message stringlengths 3 64.3k | files listlengths 1 300 | timestamp timestamp[s]date 2013-03-13 20:45:00 2026-04-11 07:48:46 |
|---|---|---|---|---|---|
ollama/ollama | cf29bd2d72f70170ef7b5adf6d3e30ac6a23331c | 3773fb646521b474d5a869d79f932a2484eb674c | fix: request retry with error
this fixes a subtle bug with makeRequestWithRetry where an HTTP status
error on a retried request will potentially not return the right err | [
{
"path": "server/images.go",
"patch": "@@ -1132,49 +1132,46 @@ func GetSHA256Digest(r io.Reader) (string, int64) {\n var errUnauthorized = fmt.Errorf(\"unauthorized\")\n \n func makeRequestWithRetry(ctx context.Context, method string, requestURL *url.URL, headers http.Header, body io.ReadSeeker, regOpts *R... | 2024-01-12T21:32:24 |
denoland/deno | e945e5c74ada0a249d26d89dc0a71dd024333b7f | cad67da32d917bf1437b0f66ebdaee07f815a384 | ci: fix saving wpt cache (#32228) | [
{
"path": ".github/workflows/ci.generate.ts",
"patch": "@@ -1171,6 +1171,15 @@ const buildJobs = buildItems.map((rawBuildItem) => {\n }\n \n if (buildItem.wpt.isPossiblyTrue()) {\n+ const buildCacheSteps = createRestoreAndSaveCacheSteps({\n+ name: \"wpt and autobahn test run hashes\",\n+ pa... | 2026-02-19T00:41:18 |
vuejs/vue | f40da5dbf2f01b417a7487d15b43cfd3aeb02a8f | 5091e2c9847601e329ac36d17eae90bb5cb77a91 | fix(v-model) selected option not updated properly under IE11 (#6213)
fix #6209 | [
{
"path": "src/platforms/web/runtime/directives/model.js",
"patch": "@@ -22,14 +22,7 @@ if (isIE9) {\n export default {\n inserted (el, binding, vnode) {\n if (vnode.tag === 'select') {\n- const cb = () => {\n- setSelected(el, binding, vnode.context)\n- }\n- cb()\n- /* istan... | 2017-08-29T15:17:38 |
ggml-org/llama.cpp | 4902eebe33ed2341ef9bd7c80195bbd9d24f4d5f | 923ae3c61983e60a2324ae56b412be5b8b511a53 | models : Added support for RND1 Diffusion Language Model (#17433)
* Converted RND1 model to GGUF weights
* RND1 llama.cpp support v1
* RND1 llama.cpp support v2 non causal bug
* RND1 llama.cpp support v3 doccumentation
* RND1 llama.cpp support v4 clean code
* linting issues
* RND1 pr fixes v1
* RND1 pr fixes v... | [
{
"path": "convert_hf_to_gguf.py",
"patch": "@@ -4183,6 +4183,21 @@ def set_vocab(self):\n super().set_vocab()\n \n \n+@ModelBase.register(\"RND1\")\n+class RND1Model(Qwen2MoeModel):\n+ model_arch = gguf.MODEL_ARCH.RND1\n+\n+ def set_gguf_parameters(self):\n+ super().set_gguf_parameters... | 2025-11-24T06:16:56 |
denoland/deno | ec11e5416eaca5c3b3db3ff836a6398c427eb83a | 409b1d4ffadcca8f67f2b424f49f8b67c4c0cfae | fix(ext/node): preserve streaming body on stale keepalive retry (#32215)
Streaming writes (e.g. pipeline, node-fetch) bypass outputData and go
directly to _bodyWriter, so they were lost when a stale keepalive
connection triggered a retry. This caused timeouts for libraries like
node-fetch and gaxios that use pipeline(... | [
{
"path": "ext/node/polyfills/_http_outgoing.ts",
"patch": "@@ -530,7 +530,13 @@ Object.defineProperties(\n }\n \n const { data, encoding, callback } = this.outputData.shift();\n- const ret = this._writeRaw(data, encoding, callback);\n+ this._flushingBuffer = true;\n+ let ret;\n+ ... | 2026-02-18T13:03:10 |
huggingface/transformers | 47227f477532e8ff6dd3290dc6b9e0e38aef303b | 7f9f4d9cc63a3e8a03d87ae178eb946cf8879fc1 | Add prefix sharing to continuous batching (#42094)
* Fix a bug in the CB memory calcuation
* Nit in example
* Replace _free_blocks with a proper object BlockManager
* Removed dead code
* Added hasing mechanism (wip)
* Added de-duplication
* Add de-initialization mechnaism
* Add prefix detection
* Ensure we alw... | [
{
"path": "examples/pytorch/continuous_batching.py",
"patch": "@@ -17,6 +17,7 @@\n import json\n import os\n import time\n+from itertools import cycle\n from typing import Optional\n \n import datasets\n@@ -29,42 +30,32 @@\n from transformers.generation.continuous_batching.requests import logger\n \n \n-# M... | 2025-11-17T12:20:15 |
ggml-org/llama.cpp | fcb013847c2c983967e9d8c9a13b16829fb799e6 | d5bc1ad11062563faa3e57320399c0938c4803fa | ggml-hexagon: Initial Hexagon v68/v69 support (#17394)
* ggml-hexagon: fix build error with GCC
Add stdexcept include to fix GCC build errors
Signed-off-by: Mohamed Mediouni <mohamed@unpredictable.fr>
* ggml-hexagon: check VTCM acquire failures
Signed-off-by: Mohamed Mediouni <mohamed@unpredictable.fr>
* ggml-he... | [
{
"path": "ggml/src/ggml-hexagon/CMakeLists.txt",
"patch": "@@ -43,6 +43,14 @@ set(HTP_CMAKE_ARGS\n -DHEXAGON_TOOLS_ROOT=$ENV{HEXAGON_TOOLS_ROOT}\n -DHEXAGON_HTP_DEBUG=${GGML_HEXAGON_HTP_DEBUG})\n \n+ExternalProject_Add(htp-v68\n+ SOURCE_DIR ${CMAKE_CURRENT_SOURCE_DIR}/htp BUILD_ALWAYS ON\n+ C... | 2025-11-24T00:54:49 |
denoland/deno | 409b1d4ffadcca8f67f2b424f49f8b67c4c0cfae | d590fb60f6e1f1069c392bf69ec55faaf9da160e | fix(audit): properly handle npm registries in subpaths (#32164)
Fixes #32106 | [
{
"path": "cli/tools/pm/audit.rs",
"patch": "@@ -149,7 +149,7 @@ mod npm {\n npm_url: Url,\n body: serde_json::Value,\n ) -> Result<AuditResponse, AnyError> {\n- let url = npm_url.join(\"/-/npm/v1/security/audits\").unwrap();\n+ let url = npm_url.join(\"-/npm/v1/security/audits\").unwrap();\... | 2026-02-18T13:01:11 |
vuejs/vue | 5091e2c9847601e329ac36d17eae90bb5cb77a91 | 0dc27dcdec72c1c2e12fb49fb95dceca45e84115 | fix(ssr): address possible xss vector | [
{
"path": "src/platforms/web/server/modules/attrs.js",
"patch": "@@ -50,7 +50,7 @@ export function renderAttr (key: string, value: string): string {\n } else if (isEnumeratedAttr(key)) {\n return ` ${key}=\"${isFalsyAttrValue(value) || value === 'false' ? 'false' : 'true'}\"`\n } else if (!isFalsyAt... | 2017-08-29T14:42:54 |
huggingface/transformers | 7f9f4d9cc63a3e8a03d87ae178eb946cf8879fc1 | 462beff5c3400ac4e5ae69eebb82bb10995d91d2 | Fix TP plans for MoE models (#42236)
* start
* more fixes | [
{
"path": "src/transformers/models/deepseek_v2/configuration_deepseek_v2.py",
"patch": "@@ -127,8 +127,9 @@ class DeepseekV2Config(PreTrainedConfig):\n \"layers.*.self_attn.q_b_proj\": \"colwise\",\n \"layers.*.self_attn.kv_b_proj\": \"colwise\",\n \"layers.*.self_attn.o_proj\": \"ro... | 2025-11-17T11:43:59 |
ggml-org/llama.cpp | d5bc1ad11062563faa3e57320399c0938c4803fa | 0c7220db56525d40177fcce3baa0d083448ec813 | ggml-hexagon: add `hex_supported_buffer` for better buffer supported check (#17212)
* hexagon: add buffer support checks for hexagon sessions
* refactor: simplify buffer support checks in hexagon operations
* hexagon: update buffer support checks to use tensor structure
* refactor: streamline buffer initialization ... | [
{
"path": "ggml/src/ggml-hexagon/ggml-hexagon.cpp",
"patch": "@@ -240,6 +240,23 @@ struct ggml_hexagon_session {\n uint32_t prof_pkts;\n };\n \n+static inline void hex_print_op_info(const ggml_tensor * op, ggml_hexagon_session * sess, const uint32_t req_flags) {\n+ char dims[64 * GGML_MAX_SRC... | 2025-11-23T22:26:36 |
denoland/deno | d590fb60f6e1f1069c392bf69ec55faaf9da160e | 476f0d9e93f0abea761c95e92eda908abb85d802 | fix(ext/node): add missing `process.constrainedMemory()` API (#32209)
## Summary
- Extracts the cgroup v1/v2 parsing logic used by
`process.constrainedMemory()` into a shared public `cgroup` module in
`ext/node/ops/process.rs`
- Reuses the shared parser from `cli/lib/worker.rs` (V8 heap limit
configuration) instead o... | [
{
"path": "cli/lib/worker.rs",
"patch": "@@ -155,6 +155,9 @@ pub fn create_isolate_create_params<TSys: DenoLibSys>(\n \n #[cfg(any(target_os = \"android\", target_os = \"linux\"))]\n mod linux {\n+ use deno_runtime::deno_node::ops::process::cgroup::CgroupVersion;\n+ use deno_runtime::deno_node::ops::proce... | 2026-02-18T11:50:11 |
vuejs/vue | 82f03de3328abde70493d2d3630137f23f140fcb | b5cc8b9e935897184ac43119bf5fbeca606f4b6a | fix-Firefox (#6359) | [
{
"path": "src/core/util/env.js",
"patch": "@@ -17,7 +17,7 @@ export const isAndroid = UA && UA.indexOf('android') > 0\n export const isIOS = UA && /iphone|ipad|ipod|ios/.test(UA)\n export const isChrome = UA && /chrome\\/\\d+/.test(UA) && !isEdge\n \n-// Firefix has a \"watch\" function on Object.prototype... | 2017-08-13T12:53:17 |
huggingface/transformers | 462beff5c3400ac4e5ae69eebb82bb10995d91d2 | 66d57110f089789ae285cc9d54d3bf051123246b | Fix initialization guard for pytest (#42234)
add copy | [
{
"path": "src/transformers/initialization.py",
"patch": "@@ -179,7 +179,7 @@ def guard_torch_init_functions():\n # something like `from torch.nn.init import xavier_uniform_` in their internals (e.g in torch.nn.modules,\n # where MultiHeadAttention lives), so the function name is bin... | 2025-11-17T10:07:16 |
denoland/deno | 951455ed1774e275b20e223e4bf3e649f6fb76a2 | f01347e771191b0aa486e9a057c981d933c3f861 | fix(bundle): let esbuild handle data urls (#32213)
Fixes #32206.
Instead of resolving them (which forces us to load them), just let
esbuild handle them | [
{
"path": "cli/tools/bundle/mod.rs",
"patch": "@@ -1235,6 +1235,9 @@ impl DenoPluginHandler {\n kind,\n with\n );\n+ if path.starts_with(\"data:\") {\n+ return Ok(None);\n+ }\n let mut resolve_dir = resolve_dir.unwrap_or(\"\").to_string();\n let resolver = self.resolver.cl... | 2026-02-17T23:23:19 |
ggml-org/llama.cpp | 028f93ef9819d1a039f97d74d72380c986cd69aa | 8e9ddba610e8c86361e068117a1935cde94b8903 | HIP: RDNA4 tensor core support for MMF (#17077)
* mmf for rdna4
* align the padding for rdna4
* forbit mul_mat_f for rdna4
* fix as comment
* remove device kernels
* add constexpr for early return
* update based on review comment
* change based on the review comment
* pass compile error
* keep code consistenc... | [
{
"path": "ggml/src/ggml-cuda/common.cuh",
"patch": "@@ -224,6 +224,10 @@ static const char * cu_get_error_str(CUresult err) {\n #define AMD_MFMA_AVAILABLE\n #endif // defined(GGML_USE_HIP) && defined(CDNA) && !defined(GGML_HIP_NO_MMQ_MFMA)\n \n+#if defined(GGML_USE_HIP) && defined(RDNA4)\n+#define AMD_WMMA... | 2025-11-21T23:03:24 |
vuejs/vue | 3d14e855e422b656859d1b419af43b94320fcfce | 303655116f8ec78f3b0ac99569637ad868dfe246 | fix: checkbox v-model="array" ignore false-value (#6180)
close #6178 | [
{
"path": "src/platforms/web/compiler/directives/model.js",
"patch": "@@ -93,7 +93,7 @@ function genCheckboxModel (\n 'if(Array.isArray($$a)){' +\n `var $$v=${number ? '_n(' + valueBinding + ')' : valueBinding},` +\n '$$i=_i($$a,$$v);' +\n- `if($$c){$$i<0&&(${value}=$$a.concat($$v))... | 2017-07-21T04:00:22 |
denoland/deno | f01347e771191b0aa486e9a057c981d933c3f861 | 3550a052e718bac916576fb91ec83d559cf915b7 | ci: fix release rebuilding on prs (#32210) | [
{
"path": ".github/workflows/ci.generate.ts",
"patch": "@@ -769,8 +769,8 @@ const buildJobs = buildItems.map((rawBuildItem) => {\n sysRootStep,\n )(\n {\n+ // do this on PRs as well as main so that PRs can use the cargo build cache from main\n nam... | 2026-02-17T21:10:17 |
ollama/ollama | 7427fa138714580f74aafa9f08d7af5d16fa2fad | de2fbdec991ac52ff015818b19482fdff22e2deb | Fix up the CPU fallback selection
The memory changes and multi-variant change had some merge
glitches I missed. This fixes them so we actually get the cpu llm lib
and best variant for the given system. | [
{
"path": "gpu/gpu_darwin.go",
"patch": "@@ -34,7 +34,7 @@ func GetGPUInfo() GpuInfo {\n \tmem, _ := getCPUMem()\n \tif runtime.GOARCH == \"amd64\" {\n \t\treturn GpuInfo{\n-\t\t\tLibrary: \"default\",\n+\t\t\tLibrary: \"cpu\",\n \t\t\tVariant: GetCPUVariant(),\n \t\t\tmemInfo: mem,\n \t\t}",
"additions... | 2024-01-11T22:43:16 |
vuejs/vue | 303655116f8ec78f3b0ac99569637ad868dfe246 | fc3d7cd7a93534d76840418467f303d4b301fbcd | fix(provide/inject): merge provide properly from mixins
close #6175 | [
{
"path": "src/core/util/options.js",
"patch": "@@ -85,7 +85,7 @@ export function mergeDataOrFn (\n return function mergedDataFn () {\n return mergeData(\n typeof childVal === 'function' ? childVal.call(this) : childVal,\n- parentVal.call(this)\n+ typeof parentVal === 'functi... | 2017-07-21T01:47:52 |
huggingface/transformers | 66d57110f089789ae285cc9d54d3bf051123246b | 8598421b5120a6a57efe9f57fbfda9bfed29a4dc | GLM-V update with new processor (#42122)
* init
* update
* add
* Update video_processing_glm46v.py
* update doc
* Update modular_glm46v.py
* 2
* Update processing_glm46v.py
* 21
* Update check_repo.py
* Update check_repo.py
* Update test_processor_glm46v.py
* Update modeling_auto.py
* update
* Update glm... | [
{
"path": "docs/source/en/_toctree.yml",
"patch": "@@ -1066,6 +1066,8 @@\n title: Gemma3n\n - local: model_doc/git\n title: GIT\n+ - local: model_doc/glm46v\n+ title: Glm46V\n - local: model_doc/glm4v\n title: glm4v\n - local: model_doc/glm4v_moe",
"... | 2025-11-15T08:44:07 |
denoland/deno | 3550a052e718bac916576fb91ec83d559cf915b7 | ac8f443b94a288fb6d0ea6503d9135933ff50644 | fix(ext/node): improve process event handling and add missing APIs (#32200)
## Summary
- Freeze `process.config` and its nested objects for Node.js
compatibility (immutability)
- Add `process.availableMemory()` using
`Deno.systemMemoryInfo().available`
- Fix `process.on/off/emit/prependListener` to handle Symbol even... | [
{
"path": "ext/node/polyfills/process.ts",
"patch": "@@ -346,6 +346,10 @@ memoryUsage.rss = function (): number {\n return memoryUsage().rss;\n };\n \n+export function availableMemory(): number {\n+ return Deno.systemMemoryInfo().available;\n+}\n+\n // Returns a negative error code than can be recognized... | 2026-02-17T19:30:22 |
vuejs/vue | fc3d7cd7a93534d76840418467f303d4b301fbcd | eb9168cfc1816b53ddb1eccd6310173a37803897 | fix: work around IE/Edge bug when accessing document.activeElement from iframe
close #6157 | [
{
"path": "src/platforms/web/runtime/modules/dom-props.js",
"patch": "@@ -63,7 +63,11 @@ function shouldUpdateValue (\n function isDirty (elm: acceptValueElm, checkVal: string): boolean {\n // return true when textbox (.number and .trim) loses focus and its value is\n // not equal to the updated value\n... | 2017-07-21T01:14:43 |
denoland/deno | 7659f3c7d7e06220cf835865fa7e96558dfa4927 | a8f201f9e8552bd5ea0823ec6454ce251fedb48f | fix(ext/node): normalize trailing dot in TLS servername (#32166)
Fixes #30170
Node.js strips trailing dots from FQDN server names before TLS
certificate verification, but Deno was passing the servername with the
trailing dot directly to rustls. This caused certificate validation
failures when connecting with a traili... | [
{
"path": "ext/node/polyfills/_tls_wrap.js",
"patch": "@@ -86,8 +86,14 @@ export class TLSSocket extends net.Socket {\n constructor(socket, opts = kEmptyObject) {\n const tlsOptions = { ...opts };\n \n- const hostname = opts.servername ?? opts.host ?? socket?._host ??\n+ let hostname = opts.serv... | 2026-02-17T12:22:50 |
ollama/ollama | d2be6387c92e2c3ed3094e5a29793e5ae74a1425 | d7af35d3d06ff9422808cf744b2ffcc0b2cdeeac | fix typo | [
{
"path": "llm/payload_common.go",
"patch": "@@ -76,7 +76,7 @@ func getDynLibs(gpuInfo gpu.GpuInfo) []string {\n \t\t}\n \t}\n \n-\t// Finaly, if we didn't find any matches, LCD CPU FTW\n+\t// Finally, if we didn't find any matches, LCD CPU FTW\n \tif len(dynLibs) == 0 {\n \t\tdynLibs = []string{availableDy... | 2024-01-11T22:25:21 |
vuejs/vue | a8ac129a5876a7eeae0137bf2f1b0968d4d6ffad | 9b4dbba384bc81a99abe429476729f80cb06d19a | fix: ensure looseEqual is not dependant on key enumeration order
close #5908 | [
{
"path": "src/shared/util.js",
"patch": "@@ -248,15 +248,31 @@ export function genStaticKeys (modules: Array<ModuleOptions>): string {\n * Check if two values are loosely equal - that is,\n * if they are plain objects, do they have the same shape?\n */\n-export function looseEqual (a: mixed, b: mixed): ... | 2017-07-21T00:41:18 |
ggml-org/llama.cpp | 28175f857d21f7acb9bd060d61d8b073219fe59c | 9cc4080441114bef1f2a13405a2a1f1d7b1ce6ca | cmake : add option to build and link BoringSSL (#17205)
* cmake: add option to build and link BoringSSL
Signed-off-by: Adrien Gallouët <angt@huggingface.co>
* cmake : fix typo
Signed-off-by: Adrien Gallouët <angt@huggingface.co>
* cmake : disable boringssl test and asm by default
Signed-off-by: Adrien Gallouët <a... | [
{
"path": ".github/workflows/build.yml",
"patch": "@@ -69,20 +69,15 @@ jobs:\n key: macOS-latest-cmake-arm64\n evict-old-files: 1d\n \n- - name: Dependencies\n- id: depends\n- continue-on-error: true\n- run: |\n- brew update\n- brew install cur... | 2025-11-21T10:46:45 |
ollama/ollama | 3bc8b9832b3bfab9ddca01ec0ebc012303362361 | ab6be852c77064d7abeffb0b03c096aab90e95fe | fix gpu_test.go Error (same type) uint64->uint32 (#1921) | [
{
"path": "gpu/gpu_test.go",
"patch": "@@ -18,7 +18,7 @@ func TestBasicGetGPUInfo(t *testing.T) {\n \tcase \"linux\", \"windows\":\n \t\tassert.Greater(t, info.TotalMemory, uint64(0))\n \t\tassert.Greater(t, info.FreeMemory, uint64(0))\n-\t\tassert.Greater(t, info.DeviceCount, uint64(0))\n+\t\tassert.Greate... | 2024-01-11T13:22:23 |
denoland/deno | a8f201f9e8552bd5ea0823ec6454ce251fedb48f | cd14853b7f67821df990eec8d43a4c772b5019bb | fix(ext/node): use primordials in ext/node/polyfills/_fs/_fs_lchown.ts (#32195)
Towards #24236
Use primordials in `ext/node/polyfills/_fs/_fs_lchown.ts`.
Changes:
- Remove `prefer-primordials` lint ignore directive
- Use `getValidatedPathToString` instead of
`getValidatedPath().toString()`
- Use `PromisePrototypeThe... | [
{
"path": "ext/node/polyfills/_fs/_fs_lchown.ts",
"patch": "@@ -1,20 +1,20 @@\n // Copyright 2018-2026 the Deno authors. MIT license.\n \n-// TODO(petamoriken): enable prefer-primordials for node polyfills\n-// deno-lint-ignore-file prefer-primordials\n-\n import {\n type CallbackWithError,\n makeCallba... | 2026-02-17T10:39:50 |
vuejs/vue | 9b4dbba384bc81a99abe429476729f80cb06d19a | 082fc3967db4d3290e901a38504dcd9bb698e561 | fix(transition): should trigger transition hooks for v-show in ie9
fix #5525 | [
{
"path": "src/platforms/web/runtime/directives/show.js",
"patch": "@@ -1,6 +1,5 @@\n /* @flow */\n \n-import { isIE9 } from 'core/util/env'\n import { enter, leave } from '../modules/transition'\n \n // recursively search for possible transition defined inside the component root\n@@ -16,7 +15,7 @@ export d... | 2017-07-21T00:03:33 |
ggml-org/llama.cpp | 21d31e0810d398f75ddd7d7c4cec9907a5576f26 | dd0f3219419b24740864b5343958a97e1b3e4b26 | ggml-hexagon: fix swiglu failure at `test-backend-ops` (#17344)
* refactor: use hvx_vec_exp_fp32_guard_inf for overflow handling in hvx_exp_f32
* feat: add fast sigmoid function with overflow guard for fp32
* refactor: replace hvx_vec_inverse_fp32 with hvx_vec_inverse_fp32_guard_inf for improved overflow handling
*... | [
{
"path": "ggml/src/ggml-hexagon/htp/act-ops.c",
"patch": "@@ -106,33 +106,32 @@ static void glu_swiglu_fp32_per_thread(const struct htp_tensor * src0,\n t1 = HAP_perf_get_qtimer_count();\n \n int is_aligned = 1;\n- int opt_path = 0;\n if (!htp_is_aligned((void *) src0->data, VLEN) || !htp_... | 2025-11-20T23:45:05 |
huggingface/transformers | 8598421b5120a6a57efe9f57fbfda9bfed29a4dc | 16c7afd06f70e1ab7a8bb2e19e33e5473297dc55 | Much more efficient and clear weight initialization and tie weights (#42191)
* everything untilo informer
* everything until perceiver
* all of them finally
* style
* replace by transformers init everywhere
* use relative import instead
* deprecated models
* style
* start contexts
* small fixes
* fix modular... | [
{
"path": "src/transformers/core_model_loading.py",
"patch": "@@ -26,7 +26,6 @@\n from contextlib import contextmanager\n from dataclasses import dataclass, field\n from functools import partial\n-from types import MethodType\n from typing import TYPE_CHECKING, Any, Optional, Union\n \n import torch\n@@ -31... | 2025-11-14T23:34:40 |
ollama/ollama | b24e8d17b29246cbf520cffc48eac374ec245e48 | f83881390f571da412d09809b6e5547783ec2392 | Increase minimum CUDA memory allocation overhead and fix minimum overhead for multi-gpu (#1896)
* increase minimum cuda overhead and fix minimum overhead for multi-gpu
* fix multi gpu overhead
* limit overhead to 10% of all gpus
* better wording
* allocate fixed amount before layers
* fixed only include... | [
{
"path": "gpu/gpu.go",
"patch": "@@ -184,10 +184,11 @@ func getCPUMem() (memInfo, error) {\n func CheckVRAM() (int64, error) {\n \tgpuInfo := GetGPUInfo()\n \tif gpuInfo.FreeMemory > 0 && (gpuInfo.Library == \"cuda\" || gpuInfo.Library == \"rocm\") {\n-\t\t// leave 10% or 384Mi of VRAM free for unaccounted... | 2024-01-11T00:08:51 |
denoland/deno | cd14853b7f67821df990eec8d43a4c772b5019bb | bdbe8bea09dd02712c03db0c28d244dad604cb44 | ci: fix mtime cache for node types (#32205)
The CI was rebuilding due to the node types directory being watched and
the mtimes cache not working with directories. | [
{
"path": "cli/build.rs",
"patch": "@@ -155,8 +155,6 @@ fn process_node_types(out_dir: &Path) {\n Ok(())\n }\n \n- println!(\"cargo:rerun-if-changed={}\", node_dir.display());\n-\n let mut paths = Vec::new();\n visit_dirs(&node_dir, &mut |path| {\n paths.push(path.to_path_buf());\n@@ -166,6 +... | 2026-02-17T04:47:47 |
vuejs/vue | 082fc3967db4d3290e901a38504dcd9bb698e561 | 17130611261fdbab70d0e5ab45036e4b612b17fe | fix(parser): only ignore the first newline in <pre>
fix #6146 | [
{
"path": "src/compiler/parser/html-parser.js",
"patch": "@@ -79,9 +79,6 @@ export function parseHTML (html, options) {\n last = html\n // Make sure we're not in a plaintext content element like script/style\n if (!lastTag || !isPlainTextElement(lastTag)) {\n- if (shouldIgnoreFirstNewline(l... | 2017-07-20T23:30:09 |
ggml-org/llama.cpp | 054a45c3d313387a4becd5eae982285932852b35 | 4c91f2633f29a51ac2dcaa1c462483ea0ef6de8a | grammar: fix regression caused by #17381 (#17412)
* grammar: fix regression caused by #17381
* more readable | [
{
"path": "src/llama-grammar.cpp",
"patch": "@@ -347,10 +347,10 @@ const char * llama_grammar_parser::parse_sequence(\n size_t last_sym_start = rule.size();\n const char * pos = src;\n \n- // use UINT64_MAX as the empty value because we aligned to the proper unsigned long type so -1 can't be used... | 2025-11-20T17:35:10 |
huggingface/transformers | 16c7afd06f70e1ab7a8bb2e19e33e5473297dc55 | 309180f93a216afee998e955440c5da9fe5ebd89 | Update `test_dynamic_cache_exportability_multiple_run` (failing on torch 2.10 nightly) (#42212)
fix
Co-authored-by: ydshieh <ydshieh@users.noreply.github.com> | [
{
"path": "tests/utils/test_cache_utils.py",
"patch": "@@ -628,7 +628,7 @@ def test_dynamic_cache_exportability_multiple_run(self):\n past_key_values = res.past_key_values\n \n shapes = torch.export.ShapesCollection()\n- dyn = torch.export.Dim(\"seq\", max=512)\n+ dyn = torch.e... | 2025-11-14T15:09:03 |
ollama/ollama | 3c49c3ab0da9cccbcca333ae278687734bb5cec8 | 9754ae4c899c5a55e7725f195f5581aad33866b2 | Harden GPU mgmt library lookup
When there are multiple management libraries installed on a system
not every one will be compatible with the current driver. This change
improves our management library algorithm to build up a set of discovered
libraries based on glob patterns, and then try all of them until we're able ... | [
{
"path": "gpu/gpu.go",
"patch": "@@ -13,7 +13,10 @@ import \"C\"\n import (\n \t\"fmt\"\n \t\"log\"\n+\t\"os\"\n+\t\"path/filepath\"\n \t\"runtime\"\n+\t\"strings\"\n \t\"sync\"\n \t\"unsafe\"\n )\n@@ -29,31 +32,79 @@ var gpuHandles *handles = nil\n // With our current CUDA compile flags, 5.2 and older wil... | 2024-01-10T22:39:51 |
denoland/deno | 25b79aab0267bbff2889c5ba1a9ba2f8112e1a95 | 8012d48ec911d2e8619a3530008496d06da01dc1 | fix(node/http): preserve request body on stale keepalive retry (#32148)
Closes https://github.com/denoland/deno/issues/32006
When a reused keepAlive socket goes stale, the retry logic was not
re-sending the request body for POST/PUT/PATCH requests. The body data
in outputData was consumed by _flushBuffer() before the... | [
{
"path": "ext/node/polyfills/http.ts",
"patch": "@@ -572,6 +572,11 @@ class ClientRequest extends OutgoingMessage {\n this._bodyWriteRid,\n baseConnRid,\n );\n+ // Save body data before flushing so it can be restored if\n+ // a stale keepAlive socket fails during t... | 2026-02-16T19:26:51 |
vuejs/vue | 17130611261fdbab70d0e5ab45036e4b612b17fe | 06b9b0bbadcc6c5afd300ed7748294e62ba00803 | fix(v-on): revert component root data.on/data.nativeOn behavior for
weex-vue-render compat
close #6109 | [
{
"path": "src/core/instance/render.js",
"patch": "@@ -54,12 +54,12 @@ export function initRender (vm: Component) {\n defineReactive(vm, '$attrs', parentData && parentData.attrs, () => {\n !isUpdatingChildComponent && warn(`$attrs is readonly.`, vm)\n }, true)\n- defineReactive(vm, '$listen... | 2017-07-19T23:39:42 |
ggml-org/llama.cpp | 92c0b387a9301ad6f9a180898d970b9cd7831716 | 2286a360ff5c6b5edd33e53b5773bdf67bc25d23 | grammar : fix integer overflow (#17381)
* Fix DoS / integer overflow
* Remove optional, use INT64_MAX instead as placeholder value (it's technically -1, so it fits :)
* White space
* Actually, since it's unsigned, use UINT64_MAX | [
{
"path": "src/llama-grammar.cpp",
"patch": "@@ -6,8 +6,10 @@\n \n #include <cmath>\n #include <algorithm>\n+#include <cstdint>\n #include <stdexcept>\n \n+#define MAX_REPETITION_THRESHOLD 2000\n //\n // helpers\n //\n@@ -345,7 +347,9 @@ const char * llama_grammar_parser::parse_sequence(\n size_t last_s... | 2025-11-20T12:47:04 |
ollama/ollama | 224fbf27959ed05f6508c3555fbd0632878a763c | 2c6e8f52484f36a39e8bc0c8d7389fd4c55e3533 | update submodule to commit `1fc2f265ff9377a37fd2c61eae9cd813a3491bea` until its main branch is fixed | [
{
"path": "llm/llama.cpp",
"patch": "@@ -1 +1 @@\n-Subproject commit 6efb8eb30e7025b168f3fda3ff83b9b386428ad6\n+Subproject commit 1fc2f265ff9377a37fd2c61eae9cd813a3491bea",
"additions": 1,
"deletions": 1,
"language": "C++"
}
] | 2024-01-10T22:03:11 |
huggingface/transformers | 309180f93a216afee998e955440c5da9fe5ebd89 | 8976ceb0510e139282050a1b12d9e6afb21bce35 | [`BLT`] Fix cache usage (#42188)
* fix
* properly
* fix tests | [
{
"path": "src/transformers/models/blt/modeling_blt.py",
"patch": "@@ -28,7 +28,7 @@\n import torch.nn.functional as F\n \n from ...activations import ACT2FN\n-from ...cache_utils import Cache, DynamicCache\n+from ...cache_utils import Cache, DynamicCache, EncoderDecoderCache\n from ...generation import Gen... | 2025-11-14T14:58:17 |
denoland/deno | 8012d48ec911d2e8619a3530008496d06da01dc1 | b248af44b755036bd0da339ae62d1a90fa049434 | fix(test): throw when Deno.test() is called during test execution (#32184)
Fixes #30787.
`Deno.test()` was being accepted even when called from inside a running
test callback.
That means nested registrations could silently be ignored and tests
looked like they passed.
Now `Deno.test()` calls made from inside a runni... | [
{
"path": "cli/ops/testing.rs",
"patch": "@@ -132,7 +132,7 @@ fn op_register_test(\n };\n state\n .borrow_mut::<TestContainer>()\n- .register(description, function);\n+ .register(description, function)?;\n ret_buf.copy_from_slice(&(id as u32).to_le_bytes());\n Ok(())\n }",
"additions":... | 2026-02-16T17:50:06 |
ggml-org/llama.cpp | 1d321e592bee0f83b0a248db9b5a8241745f25d1 | 196f5083efe636ceaf247aa4dca5593c6c2b743f | metal : fix compile on macos 11 (whisper/3533) | [
{
"path": "ggml/src/ggml-metal/ggml-metal-ops.cpp",
"patch": "@@ -11,6 +11,7 @@\n #include <cassert>\n #include <algorithm>\n #include <limits>\n+#include <cmath>\n \n static ggml_metal_buffer_id ggml_metal_get_buffer_id(const ggml_tensor * t) {\n if (!t) {",
"additions": 1,
"deletions": 0,
... | 2025-11-20T11:54:54 |
vuejs/vue | 06b9b0bbadcc6c5afd300ed7748294e62ba00803 | d03fa26687605a43d9a0c3f395d1d32375f7eaee | fix(v-bind): respect .prop modifier on components (#6159) | [
{
"path": "src/compiler/parser/index.js",
"patch": "@@ -483,8 +483,8 @@ function processAttrs (el) {\n )\n }\n }\n- if (!el.component && (\n- isProp || platformMustUseProp(el.tag, el.attrsMap.type, name)\n+ if (isProp || (\n+ !el.component && pla... | 2017-07-19T22:11:51 |
ollama/ollama | e868c8a5c794cd71704cfaeae47488438a364533 | c336693f07b77f412b2762f5327bc44eb16dc7de | Update api.md (#1878)
Fixed assistant in the example response. | [
{
"path": "docs/api.md",
"patch": "@@ -409,7 +409,7 @@ A stream of JSON objects is returned:\n \"model\": \"llama2\",\n \"created_at\": \"2023-08-04T08:52:19.385406455-07:00\",\n \"message\": {\n- \"role\": \"assisant\",\n+ \"role\": \"assistant\",\n \"content\": \"The\",\n \"images\": n... | 2024-01-09T21:21:17 |
huggingface/transformers | 8976ceb0510e139282050a1b12d9e6afb21bce35 | c01e711ee57e70af4fbf2d056411503ce49c80ec | Refactor check_auto_docstring using AST (#41432)
* refactor check_auto_docstring with AST
* use dataclass for ASTIndexes
* simplify and improve readability
* fix missing imports
* fix modular
* fix modular issues | [
{
"path": "src/transformers/models/glm4v/modeling_glm4v.py",
"patch": "@@ -1418,14 +1418,11 @@ def forward(\n pixel_values_videos: Optional[torch.FloatTensor] = None,\n image_grid_thw: Optional[torch.LongTensor] = None,\n video_grid_thw: Optional[torch.LongTensor] = None,\n- r... | 2025-11-14T14:57:08 |
denoland/deno | b248af44b755036bd0da339ae62d1a90fa049434 | 8dc2fbf7884b8a08f144aba73259306595ff9b4e | fix(ext/node): export File class from node:buffer module (#32182)
Node.js 20+ exports the `File` class from the `buffer` module. Currently
Deno's `node:buffer` polyfill exports `Blob` but not `File`, which
causes runtime errors like:
```
Error: Attempt to export a nullable value for "File"
```
This breaks frameworks... | [
{
"path": "ext/node/polyfills/buffer.ts",
"patch": "@@ -7,6 +7,7 @@ export {\n Buffer,\n constants,\n default,\n+ File,\n INSPECT_MAX_BYTES,\n isAscii,\n isUtf8,",
"additions": 1,
"deletions": 0,
"language": "Unknown"
},
{
"path": "ext/node/polyfills/internal/buffer.d.ts",
... | 2026-02-16T17:46:23 |
ggml-org/llama.cpp | 196f5083efe636ceaf247aa4dca5593c6c2b743f | 5088b435d47614b03d8e05d31f4dc693beb208ff | common : more accurate sampling timing (#17382)
* common : more accurate sampling timing
* eval-callback : minor fixes
* cont : add time_meas impl
* cont : fix log msg [no ci]
* cont : fix multiple definitions of time_meas
* llama-cli : exclude chat template init from time measurement
* cont : print percentage o... | [
{
"path": "common/common.cpp",
"patch": "@@ -26,7 +26,6 @@\n #include <sstream>\n #include <string>\n #include <thread>\n-#include <unordered_map>\n #include <unordered_set>\n #include <vector>\n \n@@ -60,6 +59,14 @@\n #pragma warning(disable: 4244 4267) // possible loss of data\n #endif\n \n+common_time_me... | 2025-11-20T11:40:10 |
vuejs/vue | d03fa26687605a43d9a0c3f395d1d32375f7eaee | b0f00e31e7d06edfdc733e2e7f24d5ca448759f9 | fix(compile): properly generate comments with special character (#6156)
close #6150 | [
{
"path": "src/compiler/codegen/index.js",
"patch": "@@ -442,7 +442,7 @@ export function genText (text: ASTText | ASTExpression): string {\n }\n \n export function genComment (comment: ASTText): string {\n- return `_e('${comment.text}')`\n+ return `_e(${JSON.stringify(comment.text)})`\n }\n \n function ge... | 2017-07-19T22:09:21 |
ollama/ollama | 1961a81f03b474eab428e30e5c9e867548fa9110 | 8a8c7e7f8dcf4b71601e9912e6dbc12064d77a24 | Set corret CUDA minimum compute capability version
If you attempt to run the current CUDA build on compute capability 5.2
cards, you'll hit the following failure:
cuBLAS error 15 at ggml-cuda.cu:7956: the requested functionality is not supported | [
{
"path": "gpu/gpu.go",
"patch": "@@ -26,8 +26,8 @@ type handles struct {\n var gpuMutex sync.Mutex\n var gpuHandles *handles = nil\n \n-// TODO verify this is the correct min version\n-const CudaComputeMajorMin = 5\n+// With our current CUDA compile flags, 5.2 and older will not work properly\n+const CudaC... | 2024-01-09T19:28:24 |
huggingface/transformers | c01e711ee57e70af4fbf2d056411503ce49c80ec | 082e3ff4a3cfaeffac38812a3a68e9a59e4d0a5a | Stop inheriting tests! (#42192)
* Stop inheriting tests!
* Just use a del instead
* fixup | [
{
"path": "tests/models/cohere2/test_modeling_cohere2.py",
"patch": "@@ -269,3 +269,6 @@ def test_generation_beyond_sliding_window(self, attn_implementation: str):\n output_text = tokenizer.batch_decode(out)\n \n self.assertEqual(output_text, EXPECTED_COMPLETIONS)\n+\n+\n+del CohereModelTest... | 2025-11-14T14:07:42 |
ggml-org/llama.cpp | 5088b435d47614b03d8e05d31f4dc693beb208ff | 845f200b2895f7b3d88c6a2503879f7b12f1eb10 | convert : fix TypeError when loading base model remotely in convert_lora_to_gguf (#17385)
* fix: TypeError when loading base model remotely in convert_lora_to_gguf
* refactor: simplify base model loading using cache_dir from HuggingFace
* Update convert_lora_to_gguf.py
Co-authored-by: Sigbjørn Skjæret <sigbjorn.skj... | [
{
"path": "convert_lora_to_gguf.py",
"patch": "@@ -277,10 +277,15 @@ def parse_args() -> argparse.Namespace:\n return parser.parse_args()\n \n \n-def load_hparams_from_hf(hf_model_id: str) -> dict[str, Any]:\n+def load_hparams_from_hf(hf_model_id: str) -> tuple[dict[str, Any], Path | None]:\n+ from h... | 2025-11-20T11:30:12 |
denoland/deno | 02c6d3d30142236d75c1de969dd342ec8c2f6f3b | 48c89eea2405f7df71f00f76ab839ba955826012 | fix(ext/node): `assert.ok` compatibility (#32173)
Allows
https://github.com/nodejs/node/blob/v24.12.0/test/parallel/test-assert-first-line.js
to pass
---------
Co-authored-by: Bartek Iwańczuk <biwanczuk@gmail.com> | [
{
"path": "Cargo.lock",
"patch": "@@ -2512,6 +2512,7 @@ dependencies = [\n \"brotli\",\n \"bytes\",\n \"data-encoding\",\n+ \"deno_ast\",\n \"deno_core\",\n \"deno_error\",\n \"deno_fetch\",",
"additions": 1,
"deletions": 0,
"language": "Unknown"
},
{
"path": "ext/node/Cargo.toml",... | 2026-02-16T14:58:54 |
ollama/ollama | f95d2f25f35c5140acbec44aaf8078c3c0526a7b | 2b9892a808a56f9d2634f85907fdc5e0498000c6 | fix temporary history file permissions | [
{
"path": "readline/history.go",
"patch": "@@ -132,7 +132,7 @@ func (h *History) Save() error {\n \n \ttmpFile := h.Filename + \".tmp\"\n \n-\tf, err := os.OpenFile(tmpFile, os.O_CREATE|os.O_WRONLY|os.O_TRUNC|os.O_APPEND, 0o666)\n+\tf, err := os.OpenFile(tmpFile, os.O_CREATE|os.O_WRONLY|os.O_TRUNC|os.O_APPE... | 2023-12-18T18:53:51 |
vuejs/vue | b0f00e31e7d06edfdc733e2e7f24d5ca448759f9 | c70addf7d1a8e820ed80b6ab14aace5aa7b604c5 | fix(provide/inject): resolve inject properly from mixins (#6107)
fix #6093 | [
{
"path": "src/core/instance/inject.js",
"patch": "@@ -1,7 +1,6 @@\n /* @flow */\n \n import { warn } from '../util/index'\n-import { hasOwn } from 'shared/util'\n import { hasSymbol } from 'core/util/env'\n import { defineReactive, observerState } from '../observer/index'\n \n@@ -56,7 +55,7 @@ export funct... | 2017-07-19T22:07:33 |
huggingface/transformers | f78cadfc974469bc4d834fca73f6cafd75ee9473 | eddd51ec3d01232572b6ef7ca02f8a15c42c7839 | [`Pop2Piano`] Fix tied weights (#42193)
* fix
* try oh try
* change fix | [
{
"path": "src/transformers/models/pop2piano/configuration_pop2piano.py",
"patch": "@@ -122,6 +122,7 @@ def __init__(\n is_encoder_decoder=is_encoder_decoder,\n **kwargs,\n )\n+ self.tie_encoder_decoder = True # forcing it\n \n \n __all__ = [\"Pop2PianoConfig\"]",
... | 2025-11-14T11:54:33 |
ggml-org/llama.cpp | 845f200b2895f7b3d88c6a2503879f7b12f1eb10 | a7784a8b1d8ceba27767bf48c92723ab3127c8a4 | ggml : Fix transposed SOLVE_TRI result (#17323)
* Did someone transpose the SOLVE_TRI result matrix? Perhaps...
* Update ggml/src/ggml-cpu/ops.cpp
Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@scala.com>
* Update ggml/src/ggml-cpu/ops.cpp
Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@scala.com>
------... | [
{
"path": "ggml/src/ggml-cpu/ops.cpp",
"patch": "@@ -9696,13 +9696,12 @@ static void ggml_compute_forward_solve_tri_f32(const struct ggml_compute_params\n for (int64_t i00 = 0; i00 < n; ++i00) {\n float sum = 0.0f;\n for (int64_t t = 0; t < i00; ++t) {\n- sum +... | 2025-11-20T10:58:21 |
denoland/deno | 5c8a33e3cffbae7437f44dabf54cc44b410d437f | 0f2fecf35ce2beb82c8a23f154ac4a395ae4ce83 | ci: fix mtime cache (#32193) | [
{
"path": ".github/workflows/ci.generate.ts",
"patch": "@@ -16,7 +16,7 @@ import {\n // Bump this number when you want to purge the cache.\n // Note: the tools/release/01_bump_crate_versions.ts script will update this version\n // automatically via regex, so ensure that this line maintains this format.\n-co... | 2026-02-16T05:40:48 |
vuejs/vue | c70addf7d1a8e820ed80b6ab14aace5aa7b604c5 | be3dc9c6e923248bcf81eb8240dd4f3c168fac59 | fix(v-model): use stricter check for <select> option update
close #6112 | [
{
"path": "src/platforms/web/runtime/directives/model.js",
"patch": "@@ -30,6 +30,7 @@ export default {\n if (isIE || isEdge) {\n setTimeout(cb, 0)\n }\n+ el._vOptions = [].map.call(el.options, getValue)\n } else if (vnode.tag === 'textarea' || isTextInputType(el.type)) {\n ... | 2017-07-19T21:51:41 |
ollama/ollama | 2b9892a808a56f9d2634f85907fdc5e0498000c6 | 2bb2bdd5d41af52a19d34cf3ee5d4148839562e5 | fix(windows): modelpath and list | [
{
"path": "server/modelpath.go",
"patch": "@@ -46,7 +46,7 @@ func ParseModelPath(name string) ModelPath {\n \t\tname = after\n \t}\n \n-\tparts := strings.Split(name, string(os.PathSeparator))\n+\tparts := strings.Split(name, \"/\")\n \tswitch len(parts) {\n \tcase 3:\n \t\tmp.Registry = parts[0]",
"add... | 2023-12-15T23:50:51 |
huggingface/transformers | eddd51ec3d01232572b6ef7ca02f8a15c42c7839 | 7607d80f7e584328e0e37ca93e661998aee69a4d | Fix checkpoint loading with DeepSpeed ZeRO3 (#42201)
fix checkpoint loading with DeepSpeed ZeRO3
Signed-off-by: Masahiro Tanaka <mtanaka@anyscale.com>
Co-authored-by: Ferdinand Mom <47445085+3outeille@users.noreply.github.com> | [
{
"path": "src/transformers/modeling_utils.py",
"patch": "@@ -4233,6 +4233,19 @@ def _load_pretrained_model(\n error_msgs = []\n \n if is_deepspeed_zero3_enabled() and not is_quantized:\n+ if state_dict is None:\n+ if checkpoint_files is None:\n+ ... | 2025-11-14T10:48:08 |
ggml-org/llama.cpp | 79bb74351213da6c368fd53372be17378458b161 | 3ae282a06f74f8a24fa1ac3b1ab6dc8dda4ea22c | ggml : remove useless and error-prone variadic macros (#17399)
Signed-off-by: Adrien Gallouët <angt@huggingface.co> | [
{
"path": "ggml/src/ggml-cpu/simd-mappings.h",
"patch": "@@ -160,18 +160,18 @@ inline static float ggml_lookup_fp16_to_fp32(ggml_fp16_t f) {\n #define GGML_F32xt svfloat32_t\n #define GGML_F32xt_ZERO svdup_n_f32(0.0f)\n #define GGML_F32xt_SET1(x) svdup... | 2025-11-20T10:18:27 |
vuejs/vue | be3dc9c6e923248bcf81eb8240dd4f3c168fac59 | 381b48503aae796c5c21c318cae8b0a8322dde35 | fix: include boolean in isPrimitive check (#6127)
suppresses key warning for boolean values, closes #6126 | [
{
"path": "src/shared/util.js",
"patch": "@@ -22,7 +22,11 @@ export function isFalse (v: any): boolean %checks {\n * Check if value is primitive\n */\n export function isPrimitive (value: any): boolean %checks {\n- return typeof value === 'string' || typeof value === 'number'\n+ return (\n+ typeof va... | 2017-07-19T12:27:48 |
ollama/ollama | 2bb2bdd5d41af52a19d34cf3ee5d4148839562e5 | acfc376efda39919baeca9a8b68992b7fb74bcc0 | fix lint | [
{
"path": "cmd/interactive.go",
"patch": "@@ -238,10 +238,7 @@ func generateInteractive(cmd *cobra.Command, opts generateOptions) error {\n \t\t\t\t\t\tusageParameters()\n \t\t\t\t\t\tcontinue\n \t\t\t\t\t}\n-\t\t\t\t\tvar params []string\n-\t\t\t\t\tfor _, p := range args[3:] {\n-\t\t\t\t\t\tparams = appen... | 2023-12-15T22:07:34 |
denoland/deno | 1951149659c32b26c9f1dfb557eecc20a278e298 | 21461bd0bf6cb3105e10f07e349ed07ede9a6f34 | fix(ext/node): process.exit() in worker immediately halts execution (#32169)
## Summary
In Node.js, `process.exit()` inside a worker immediately stops all JS
execution at the C++ level. In Deno, it called `workerClose()` which
triggered V8's `terminate_execution()`, but JS continued running because
V8 only checks the... | [
{
"path": "ext/node/polyfills/process.ts",
"patch": "@@ -128,6 +128,21 @@ export const exit = (code?: number | string) => {\n // Any valid thing `process.exitCode` set is already held in Deno.exitCode.\n // At this point, we don't have to pass around Node's raw/string exit value.\n process.reallyExit(... | 2026-02-15T21:18:35 |
huggingface/transformers | 32a58e31463e238c967207bf73772490c353551a | 6f6095e0cf509f7384d3ce0c1804013ef6cafd5f | 🚨 Delete deprecations with end-cycle in v4.xx and v5.0 (#41681)
* remove deprecations from v4
* delete those for v5
* delete these also
* fix tests
* add dummy test config
* fix copies
* SDPA raises warning but doesn't automatically change to eager
* max size can't be deleted, sadly
* oke, this should allow l... | [
{
"path": "docs/source/en/model_doc/qwen2_5_omni.md",
"patch": "@@ -136,7 +136,7 @@ inputs = processor.apply_chat_template(\n tokenize=True,\n return_dict=True,\n return_tensors=\"pt\",\n- video_fps=1,\n+ fps=1,\n \n # kwargs to be passed to `Qwen2-5-OmniProcessor`\n padding=True,\... | 2025-11-14T09:26:16 |
denoland/deno | 21461bd0bf6cb3105e10f07e349ed07ede9a6f34 | bc1e7e2b203362296cb24268fb7c07f47c3aeafc | chore: fix deno_features rebuilding when unnecessary (#32186) | [
{
"path": "runtime/features/build.rs",
"patch": "@@ -8,6 +8,12 @@ mod data;\n mod structs;\n \n fn main() {\n+ // only rerun if the build script or its input modules change,\n+ // not when any file in the package changes (the default)\n+ println!(\"cargo:rerun-if-changed=build.rs\");\n+ println!(\"cargo... | 2026-02-15T21:04:41 |
ggml-org/llama.cpp | 3ae282a06f74f8a24fa1ac3b1ab6dc8dda4ea22c | 5be353ec4a4c74a9efc96ba686264bf4a282a493 | kleidiai: fix zero-size array declaration (#17240) | [
{
"path": "ggml/src/ggml-cpu/kleidiai/kernels.cpp",
"patch": "@@ -39,7 +39,7 @@\n \n #include \"kernels.h\"\n \n-#define NELEMS(x) sizeof(x) / sizeof(*x)\n+#define NELEMS(x) (sizeof(x) / sizeof(*x))\n \n template<size_t(*Fn)(size_t,size_t,size_t)>\n static inline size_t kernel_offs_fn3(size_t a, size_t b, s... | 2025-11-20T09:45:49 |
ollama/ollama | 37708931fb37f228a9bf5c74c4c98ad1b1c9650e | f6cb0a553cc2ab87192c5ac9f001ad7c366d396e | update cuda overhead to 20% to fix crashes when switching between models and large context sizes | [
{
"path": "gpu/gpu.go",
"patch": "@@ -131,14 +131,8 @@ func getCPUMem() (memInfo, error) {\n func CheckVRAM() (int64, error) {\n \tgpuInfo := GetGPUInfo()\n \tif gpuInfo.FreeMemory > 0 && (gpuInfo.Library == \"cuda\" || gpuInfo.Library == \"rocm\") {\n-\t\t// leave 15% or 400MiB of VRAM free for overhead\n-... | 2024-01-09T05:05:23 |
denoland/deno | afa1646de07c4a60f7eb94c58362d4b01f312477 | ad9c6cac029a095d5fb79e3b2d8fa0aac0dc7bfd | fix(ext/node): `fs.rmdir` compatibility (#32144)
Closes #32116
---------
Co-authored-by: Bartek Iwańczuk <biwanczuk@gmail.com> | [
{
"path": "cli/rt/file_system.rs",
"patch": "@@ -269,6 +269,15 @@ impl FileSystem for DenoRtSys {\n RealFs.remove_async(path, recursive).await\n }\n \n+ fn rmdir_sync(&self, path: &CheckedPath) -> FsResult<()> {\n+ self.error_if_in_vfs(path)?;\n+ RealFs.rmdir_sync(path)\n+ }\n+ async fn rmdir... | 2026-02-15T13:11:00 |
ggml-org/llama.cpp | 5be353ec4a4c74a9efc96ba686264bf4a282a493 | 7d77f07325985c03a91fa371d0a68ef88a91ec7f | ggml-cpu:add RISC-V RVV (Zvfh) optimization for FP16 vector scaling (#17314)
* ggml-cpu:add RISC-V RVV (Zvfh) optimization for FP16 vector scaling
Signed-off-by: Wang Yang <yangwang@iscas.ac.cn>
* fix comment
* fix comment 2
---------
Signed-off-by: Wang Yang <yangwang@iscas.ac.cn> | [
{
"path": "ggml/src/ggml-cpu/vec.h",
"patch": "@@ -698,60 +698,61 @@ inline static void ggml_vec_scale_f32(const int n, float * y, const float v) {\n }\n \n inline static void ggml_vec_scale_f16(const int n, ggml_fp16_t * y, const float v) {\n-#if defined(GGML_SIMD)\n- #if defined(__ARM_FEATURE_SVE)\n-... | 2025-11-20T06:09:18 |
vuejs/vue | a7615ef053d8bca85623666cb9a68e4d21f76952 | ac3d1eaea34ee23978d395a84f68154e7e86caa1 | build: fix postinstall script causing user install errors | [
{
"path": ".github/CONTRIBUTING.md",
"patch": "@@ -41,9 +41,10 @@ After cloning the repo, run:\n \n ``` bash\n $ npm install\n+& npm run setup\n ```\n \n-This will also run the `postinstall` script which links two git hooks:\n+The `setup` script links two git hooks:\n \n - `pre-commit`: runs ESLint on stage... | 2017-07-13T06:38:31 |
denoland/deno | ad9c6cac029a095d5fb79e3b2d8fa0aac0dc7bfd | 3127eab864feedc4f88bef6b97210581d174c1e6 | fix(ext/node): support O_RDONLY | O_CREAT in fs.openSync (#32152)
## Description
POSIX allows `O_RDONLY | O_CREAT` to create a file if it doesn't exist
and then open it for reading only. Node.js and other runtimes support
this combination. However, Rust's `std::fs::OpenOptions` requires write
or append access when `c... | [
{
"path": "ext/fs/std_fs.rs",
"patch": "@@ -1029,13 +1029,49 @@ fn open_options(options: OpenOptions) -> fs::OpenOptions {\n open_options\n }\n \n-#[inline(always)]\n pub fn open_with_checked_path(\n- options: OpenOptions,\n+ opts: OpenOptions,\n path: &CheckedPath,\n ) -> FsResult<std::fs::File> {\n-... | 2026-02-15T13:10:48 |
ollama/ollama | 2680078c137a93e12a55392a7e4d64632c58fb7b | f1b7e5f560e5c45b13f4c9dfb8114d67bb99d3d9 | fix build on linux | [
{
"path": "gpu/gpu.go",
"patch": "@@ -132,7 +132,7 @@ func CheckVRAM() (int64, error) {\n \tgpuInfo := GetGPUInfo()\n \tif gpuInfo.FreeMemory > 0 && (gpuInfo.Library == \"cuda\" || gpuInfo.Library == \"rocm\") {\n \t\t// leave 15% or 400MiB of VRAM free for overhead\n-\t\toverhead := gpuInfo.FreeMemory 3 / ... | 2024-01-09T04:44:13 |
vuejs/vue | 086e6d98f4217542afcc2794717119c62dde89b8 | dc1bca1361807596f73340e54d76d16568b3224f | fix: handle arrays in v-on object syntax | [
{
"path": "src/core/instance/render-helpers/bind-object-listeners.js",
"patch": "@@ -14,7 +14,7 @@ export function bindObjectListeners (data: any, value: any): VNodeData {\n for (const key in value) {\n const existing = on[key]\n const ours = value[key]\n- on[key] = existing ? [... | 2017-07-13T05:41:01 |
huggingface/transformers | 6f6095e0cf509f7384d3ce0c1804013ef6cafd5f | c4cfc2e023392eb75d8735c3276c3a9ba7e4ffa5 | Refactor weight loading (#41580)
* ah actually we don't discard lm head if missing -> needs to be moved to correct device and etc
* fix some tests
* small fixes
* up
* up
* dik why we tie weights twice but,..,,.
* ups
* removeunused
* fix hunyuan
* small fix
* nits
* ish
* up
* rev
* fix more tie weights... | [
{
"path": ".circleci/config.yml",
"patch": "@@ -46,8 +46,8 @@ jobs:\n - run: uv pip install -U -e .\n - run: echo 'export \"GIT_COMMIT_MESSAGE=$(git show -s --format=%s)\"' >> \"$BASH_ENV\" && source \"$BASH_ENV\"\n - run: mkdir -p test_preparation\n- - run: py... | 2025-11-13T16:12:52 |
ggml-org/llama.cpp | 99c53d6558e1882fdddd1077495b618f3210cc02 | 07b0e7a5ac5ae2ed553fccb12d452c027fb3c060 | webui: Add a "Continue" Action for Assistant Message (#16971)
* feat: Add "Continue" action for assistant messages
* feat: Continuation logic & prompt improvements
* chore: update webui build output
* feat: Improve logic for continuing the assistant message
* chore: update webui build output
* chore: Linting
* c... | [
{
"path": "tools/server/webui/.gitignore",
"patch": "@@ -25,3 +25,4 @@ vite.config.ts.timestamp-*\n \n *storybook.log\n storybook-static\n+*.code-workspace\n\\ No newline at end of file",
"additions": 1,
"deletions": 0,
"language": "Unknown"
},
{
"path": "tools/server/webui/package-lock.... | 2025-11-19T13:39:50 |
ollama/ollama | 18ddf6d57d809b22e12f276854b2949f8b52d956 | 61e650244963c791ddaf11051f69d9ff2eb49221 | fix windows build | [
{
"path": "llm/ext_server_windows.go",
"patch": "@@ -4,9 +4,9 @@ import (\n \t\"github.com/jmorganca/ollama/api\"\n )\n \n-func newDefaultExtServer(model string, adapters, projectors []string, numLayers int64, opts api.Options) (extServer, error) {\n+func newDefaultExtServer(model string, adapters, projecto... | 2024-01-09T01:04:01 |
denoland/deno | 3127eab864feedc4f88bef6b97210581d174c1e6 | 0c839b0891e84faba19a3a27c772bd75e6f232d4 | fix(ext/node): add parentPath to Dirent from fs.opendir and support Buffer paths in lstat (#32178)
Closes https://github.com/denoland/deno/issues/32175
## Summary
- **Fix `fs.opendir` Dirent `parentPath`**: `Dir.read()` and
`Dir.readSync()` now pass the directory path to `direntFromDeno()`, so
Dirents correctly have... | [
{
"path": "ext/node/polyfills/_fs/_fs_dir.ts",
"patch": "@@ -50,12 +50,16 @@ export default class Dir {\n AsyncGeneratorPrototypeNext(this.#asyncIterator),\n (iteratorResult) => {\n resolve(\n- iteratorResult.done ? null : direntFromDeno(iteratorResult.value),\n+ ... | 2026-02-15T13:05:32 |
vuejs/vue | 91deb4fd910afd51137820f120d4c26d0a99e629 | 6bf97721f1e38713353d5ac9906c88dca2cdad9b | fix: multiple merged vnode hooks not invoked properly
fix #6076 | [
{
"path": "src/core/vdom/helpers/update-listeners.js",
"patch": "@@ -27,8 +27,9 @@ export function createFnInvoker (fns: Function | Array<Function>): Function {\n function invoker () {\n const fns = invoker.fns\n if (Array.isArray(fns)) {\n- for (let i = 0; i < fns.length; i++) {\n- fn... | 2017-07-12T09:56:04 |
ggml-org/llama.cpp | fd7353d5eb828def58b1e2afc09ea01df395d8cb | 6fd4f953673e6a91041b827c45cdfd1d85e9bb64 | cuda: fix rope fusion for gemma3 (#17378) | [
{
"path": "ggml/src/ggml-cuda/ggml-cuda.cu",
"patch": "@@ -3001,6 +3001,10 @@ static void update_cuda_graph_executable(ggml_backend_cuda_context * cuda_ctx) {\n static bool ggml_cuda_should_fuse_rope_set_rows(const ggml_tensor * rope,\n const ggml_tensor * vie... | 2025-11-19T10:25:05 |
huggingface/transformers | c4cfc2e023392eb75d8735c3276c3a9ba7e4ffa5 | 5c6d6bed4d2ed5ee3380c5003f61c500049b9dc5 | [TP] Fix parameter detection issue and some invalid TP-plans (#42129)
* fix
* add test
* fix test
* fix the obvious
* more fix
* fix
* continue to improve
* more fix
* more
* fix
* fix
* finally
* CI | [
{
"path": "src/transformers/integrations/tensor_parallel.py",
"patch": "@@ -140,6 +140,16 @@ def _blocks_to_block_sizes(total_size: int, blocks: int | list[int]) -> list[int\n return [single_size] * blocks\n \n \n+def replace_layer_number_by_wildcard(name: str) -> str:\n+ \"\"\"\n+ Replace the... | 2025-11-13T14:44:56 |
denoland/deno | 0c839b0891e84faba19a3a27c772bd75e6f232d4 | e3fda850d5215d4a78e95fbc0ae86096c2f12546 | fix(ext/node): fix worker.terminate() return value and exit code (#32168)
## Summary
Two fixes to `worker.terminate()` behavior to match Node.js:
- **Return `undefined` when worker already exited**: In Node.js, calling
`worker.terminate()` after the worker has already exited resolves to
`undefined` (the internal han... | [
{
"path": "ext/node/polyfills/worker_threads.ts",
"patch": "@@ -601,7 +601,7 @@ class NodeWorker extends EventEmitter {\n // https://nodejs.org/api/worker_threads.html#workerterminate\n terminate() {\n if (this.#status === \"TERMINATED\") {\n- return PromiseResolve(0);\n+ return PromiseRes... | 2026-02-15T13:04:43 |
vuejs/vue | 606666d5a9008ab72f7cce9803b26310c636eef8 | 1f9e924971d7894517075f7f0efeeb85994a7ba0 | test: fix Object.prototype.watch related warnings | [
{
"path": "src/core/instance/state.js",
"patch": "@@ -20,6 +20,7 @@ import {\n hasOwn,\n isReserved,\n handleError,\n+ nativeWatch,\n validateProp,\n isPlainObject,\n isReservedAttribute\n@@ -53,7 +54,9 @@ export function initState (vm: Component) {\n observe(vm._data = {}, true /* asRootDa... | 2017-07-12T07:50:09 |
ggml-org/llama.cpp | 6fd4f953673e6a91041b827c45cdfd1d85e9bb64 | 980b7cd17e055c8c587f79ffda7eb4fddf405566 | Fix too relaxed check on CUDA "fast copy" (can_be_transposed) condition (#17332)
* Fix too relaxed check on CUDA "fast copy" (can_be_transposed) condition
* Argh.
* Making CISC happy ;)
* Integrate CONT tests
* Use loopy loop
* Skip new tests for (B)F16 for now. | [
{
"path": "ggml/src/ggml-cuda/cpy.cu",
"patch": "@@ -384,7 +384,8 @@ void ggml_cuda_cpy(ggml_backend_cuda_context & ctx, const ggml_tensor * src0, gg\n char * src1_ddc = (char *) src1->data;\n \n const bool contiguous_srcs = ggml_is_contiguous(src0) && ggml_is_contiguous(src1);\n- const bool can_... | 2025-11-19T09:36:33 |
huggingface/transformers | 5c6d6bed4d2ed5ee3380c5003f61c500049b9dc5 | 80134e6e663db962c30d0e45c54265b3141bc7c7 | [`PEFT`] Fix the general test for prefix tuning (#42185)
fix | [
{
"path": "tests/utils/test_modeling_utils.py",
"patch": "@@ -1818,6 +1818,8 @@ def test_cache_when_needed_at_train_time(self):\n self.assertTrue(model.training)\n \n # We can also disable the cache to skip a few operations, if the training loop doesn't need cache\n+ # NOTE: after #41... | 2025-11-13T14:40:01 |
denoland/deno | e3fda850d5215d4a78e95fbc0ae86096c2f12546 | c3fe9932581c7396d50c1b228012eb6c9102f817 | fix(ext/node): add worker_threads stdin support (#32165)
## Summary
- Adds `worker.stdin` support when `{ stdin: true }` is passed to the
`Worker` constructor
- Parent creates a `Writable` stream (`worker.stdin`) that sends
`WORKER_STDIN` / `WORKER_STDIN_END` messages to the worker via
`postMessage`
- Worker replaces... | [
{
"path": "ext/node/polyfills/worker_threads.ts",
"patch": "@@ -38,7 +38,7 @@ import {\n validateObject,\n } from \"ext:deno_node/internal/validators.mjs\";\n import { EventEmitter } from \"node:events\";\n-import { Readable } from \"node:stream\";\n+import { Readable, Writable } from \"node:stream\";\n i... | 2026-02-14T18:26:38 |
ollama/ollama | 08f1e18965c15648504fc5ec367134898e92ec6d | 7e8f7c835859b54abe50dc855f90357db27c7059 | Offload layers to GPU based on new model size estimates (#1850)
* select layers based on estimated model memory usage
* always account for scratch vram
* dont load +1 layers
* better estmation for graph alloc
* Update gpu/gpu_darwin.go
Co-authored-by: Bruce MacDonald <brucewmacdonald@gmail.com>
* Upd... | [
{
"path": "gpu/gpu.go",
"patch": "@@ -16,8 +16,6 @@ import (\n \t\"runtime\"\n \t\"sync\"\n \t\"unsafe\"\n-\n-\t\"github.com/jmorganca/ollama/api\"\n )\n \n type handles struct {\n@@ -133,31 +131,14 @@ func getCPUMem() (memInfo, error) {\n func CheckVRAM() (int64, error) {\n \tgpuInfo := GetGPUInfo()\n \tif... | 2024-01-08T21:42:00 |
vuejs/vue | 0a9aab510bcca85c78ef06487b5dcf25d76d780d | fed602b90be89a43b4ec41b5a2e0a526d3a5f4a2 | fix(v-model): fix input change check for type="number"
The change should only apply to the .number modifier, not type="number".
fix #6069 | [
{
"path": "src/platforms/web/compiler/directives/model.js",
"patch": "@@ -158,7 +158,7 @@ function genDefaultModel (\n \n addProp(el, 'value', `(${value})`)\n addHandler(el, event, code, null, true)\n- if (trim || number || type === 'number') {\n+ if (trim || number) {\n addHandler(el, 'blur', '$f... | 2017-07-11T05:04:55 |
ggml-org/llama.cpp | 980b7cd17e055c8c587f79ffda7eb4fddf405566 | c49daff5baf34cabe41b70f99cc42a9dec87c8ae | vulkan: force full subgroups for flash attention to fix intel subgroup crash (#17356) | [
{
"path": "ggml/src/ggml-vulkan/ggml-vulkan.cpp",
"patch": "@@ -2901,15 +2901,15 @@ static void ggml_vk_load_shaders(vk_device& device) {\n if (path == FAPATH) { \\\n if (aligned) { \\\n if (f32acc) { \\\n- ggml_vk_create_pipeline(device... | 2025-11-19T07:46:26 |
huggingface/transformers | 80134e6e663db962c30d0e45c54265b3141bc7c7 | ce40ca0d4c7d2e0a3f8bd3ddc30f29c6a105efb5 | Update transformers to support `FqnToConfig` (#41894)
* Update transformers to support `FqnToConfig`
Summary:
Test Plan:
Reviewers:
Subscribers:
Tasks:
Tags:
* add case for modulefqn
* remove comment
* update tests
* cleanup
* update
* wip
* wip
* update quantizer_torchao for module default
* fix unders... | [
{
"path": "docs/source/en/quantization/torchao.md",
"patch": "@@ -422,19 +422,19 @@ print(tokenizer.decode(output[0], skip_special_tokens=True))\n \n #### 1. Skip quantization for certain layers\n \n-With `ModuleFqnToConfig` we can specify a default configuration for all layers while skipping quantization f... | 2025-11-13T13:34:45 |
vuejs/vue | 4d680794a5a345078827a3fee3db8658bd35ec3a | a1d1145c9123f7175f3ac20b503cfa507ad455f4 | fix(parser): the first newline following pre and textarea tag should be ignored (#6022)
fix #6022 | [
{
"path": "src/compiler/parser/html-parser.js",
"patch": "@@ -59,6 +59,10 @@ const decodingMap = {\n const encodedAttr = /&(?:lt|gt|quot|amp);/g\n const encodedAttrWithNewLines = /&(?:lt|gt|quot|amp|#10);/g\n \n+// #5992\n+const isIgnoreNewlineTag = makeMap('pre,textarea', true)\n+const shouldIgnoreFirstNew... | 2017-07-10T13:42:00 |
ggml-org/llama.cpp | 10e9780154365b191fb43ca4830659ef12def80f | a045492088dd105231cc5f8bfaae2d527ce47344 | chat: fix int overflow, prevent size calculation in float/double (#17357)
* chat: fix int overflow, prevent size calculation in float/double
* Update common/chat.cpp
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
---------
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> | [
{
"path": "common/chat.cpp",
"patch": "@@ -3359,7 +3359,7 @@ static common_chat_params common_chat_templates_apply_legacy(\n const struct common_chat_templates * tmpls,\n const struct common_chat_templates_inputs & inputs)\n {\n- int alloc_size = 0;\n+ size_t alloc_size = 0;\n std::vector<... | 2025-11-18T18:11:53 |
huggingface/transformers | ce40ca0d4c7d2e0a3f8bd3ddc30f29c6a105efb5 | 6408d3b01a4b5446655c222daa0b947bddf7db82 | [typo] fix mrope-interleave annotation to avoid ambiguity (#42177)
* fix typo
* fix qwen3omni | [
{
"path": "src/transformers/models/qwen3_omni_moe/modeling_qwen3_omni_moe.py",
"patch": "@@ -1292,7 +1292,7 @@ def forward(self, x, position_ids):\n def apply_interleaved_mrope(self, freqs, mrope_section):\n \"\"\"Apply interleaved MRoPE to 3D rotary embeddings.\n Reorganizes frequency l... | 2025-11-13T13:07:14 |
denoland/deno | c3fe9932581c7396d50c1b228012eb6c9102f817 | 021eb21b09d85e4fb8b29f4e7a0efb78397daf28 | fix(ext/node): make worker_threads ref/unref idempotent like Node.js (#32161)
Closes #31871
## Summary
- Changes `Worker.ref()`/`Worker.unref()` from a reference counter to an
idempotent boolean toggle, matching Node.js semantics
- In Node.js, `ref()`/`unref()` are idempotent: calling `ref()` multiple
times is the sa... | [
{
"path": "ext/node/polyfills/worker_threads.ts",
"patch": "@@ -160,7 +160,7 @@ const privateWorkerRef = Symbol(\"privateWorkerRef\");\n class NodeWorker extends EventEmitter {\n #id = 0;\n #name = \"\";\n- #refCount = 1;\n+ #refed = true;\n #messagePromise = undefined;\n #controlPromise = undefin... | 2026-02-14T17:34:10 |
ollama/ollama | d74ce6bd4f78f8a06c97bf9b24485211c48a41d8 | 57942b467696f9d7dbd194a2591d144e1da596b9 | Detect very old CUDA GPUs and fall back to CPU
If we try to load the CUDA library on an old GPU, it panics and crashes
the server. This checks the compute capability before we load the
library so we can gracefully fall back to CPU mode. | [
{
"path": "gpu/gpu.go",
"patch": "@@ -28,6 +28,9 @@ type handles struct {\n var gpuMutex sync.Mutex\n var gpuHandles *handles = nil\n \n+// TODO verify this is the correct min version\n+const CudaComputeMajorMin = 5\n+\n // Note: gpuMutex must already be held\n func initGPUHandles() {\n \t// TODO - if the o... | 2024-01-07T05:40:04 |
ggml-org/llama.cpp | a045492088dd105231cc5f8bfaae2d527ce47344 | 1920345c3bcec451421bb6abc4981678cc721154 | vocab : call reserve() for building plamo-2-translate suffix (#17343)
Test 'Q4_K_M' quantization on https://huggingface.co/pfnet/plamo-2-translate
The 'suffix_to_score' size is 193510, it needs 19 memory allocation with final
capacity 262144 to hold the value, if not preserve the memory.
Signed-off-by: Haiyue Wang <... | [
{
"path": "src/llama-vocab.cpp",
"patch": "@@ -1281,6 +1281,7 @@ struct llm_tokenizer_plamo2 : llm_tokenizer {\n \n // Build suffix list in lexicographical order of reversed strings\n std::vector<std::string> suffixes;\n+ suffixes.reserve(suffix_to_score.size() + 1);\n for (co... | 2025-11-18T17:58:22 |
vuejs/vue | a1d1145c9123f7175f3ac20b503cfa507ad455f4 | 8d66691ee264969447390822971b8caa029cac3e | fix(v-model): should generate component-specific code for tags with "is" attribute
fix #6066 | [
{
"path": "src/compiler/parser/index.js",
"patch": "@@ -483,7 +483,9 @@ function processAttrs (el) {\n )\n }\n }\n- if (isProp || platformMustUseProp(el.tag, el.attrsMap.type, name)) {\n+ if (!el.component && (\n+ isProp || platformMustUseProp(el.tag, el.... | 2017-07-10T13:34:40 |
Subsets and Splits
Assembly Language GitHub Issues
Retrieves a sample of assembly-language related commits with their details, but doesn't provide meaningful analysis or patterns beyond basic filtering.
Swift Compiler Issues Analysis
Retrieves all training data for the Swift programming language repository, providing basic filtering but offering limited analytical insight beyond identifying relevant code examples.