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 |
|---|---|---|---|---|---|
denoland/deno | bad6eaeb506718d3ff990ba88821f6c3fcd1f2ab | 92d7ec6caec4fc0d8b63bc820411c048594eade4 | fix(ext/node): drain all ready connections in poll_tcp_handle to prevent accept starvation (#32943)
## Summary
- Fix `node:net` accept starvation under high concurrency when mixed
with Deno-native APIs (`Deno.listenTls()`, `Deno.connect()`)
- `poll_tcp_handle` now drains all ready connections from `poll_accept`
in a ... | [
{
"path": "libs/core/uv_compat/tcp.rs",
"patch": "@@ -606,34 +606,41 @@ pub(crate) unsafe fn poll_tcp_handle(\n }\n \n // 2. Poll listener for new connections.\n- // Match libuv: accept one connection at a time. Only poll when the\n- // backlog is empty. If the user doesn't call uv_accept in the... | 2026-03-24T12:12:19 |
huggingface/transformers | 79432f7a4c1a4dd39ebdc3622dfe2dd45eac2589 | ade62c2a0f4e8fbe2df9f00c733b6c04f112c135 | [`Auto`] Make processor subclasses overridable on load time (#42912)
fix | [
{
"path": "src/transformers/models/auto/feature_extraction_auto.py",
"patch": "@@ -348,13 +348,13 @@ def from_pretrained(cls, pretrained_model_name_or_path, **kwargs):\n )\n _ = kwargs.pop(\"code_revision\", None)\n feature_extractor_class.register_for_auto_class()\n- ... | 2025-12-17T13:15:34 |
ollama/ollama | da3bf2335483fac7cbfff72a1b80e40988f4f35a | 45cacbaf0568a4d38d74ebdd0957fe01bd06719d | Workaround gfx900 SDMA bugs
Implement support for GPU env var workarounds, and leverage
this for the Vega RX 56 which needs
HSA_ENABLE_SDMA=0 set to work properly | [
{
"path": "gpu/amd_linux.go",
"patch": "@@ -332,6 +332,11 @@ func AMDGetGPUInfo() []RocmGPUInfo {\n \t\t\tslog.Info(\"skipping rocm gfx compatibility check\", \"HSA_OVERRIDE_GFX_VERSION\", gfxOverride)\n \t\t}\n \n+\t\t// Check for env var workarounds\n+\t\tif name == \"1002:687f\" { // Vega RX 56\n+\t\t\tg... | 2024-05-31T23:15:21 |
denoland/deno | c09b7f4511abe8b156839ecfc5941178db8aa5a1 | 71ac11e15ad5ea5088458e1a0b4a8c6109a18f07 | fix(ext/web): handle null options in removeEventListener (#32605)
Fixes #20101
`normalizeEventHandlerOptions` (used by `removeEventListener`) throws
`TypeError: Cannot read properties of null` when `null` is passed as
the options parameter. Per the DOM spec, `null` should flatten to
`capture: false`.
---------
Co-a... | [
{
"path": "ext/web/02_event.js",
"patch": "@@ -834,7 +834,10 @@ function invokeEventListeners(tuple, eventImpl) {\n function normalizeEventHandlerOptions(\n options,\n ) {\n- if (typeof options === \"boolean\" || typeof options === \"undefined\") {\n+ if (\n+ typeof options === \"boolean\" || typeof ... | 2026-03-24T07:49:59 |
vuejs/vue | b00868c5cb1ca8aa8b42947ff5d2a55e9e298582 | 35edc1c1e2515033a07e903580df01bcd43090aa | refactor: extract async error handling logic | [
{
"path": "src/core/instance/events.js",
"patch": "@@ -4,9 +4,8 @@ import {\n tip,\n toArray,\n hyphenate,\n- handleError,\n formatComponentName,\n- handlePromiseError\n+ invokeWithErrorHandling\n } from '../util/index'\n import { updateListeners } from '../vdom/helpers/index'\n \n@@ -134,13 +133... | 2018-12-20T15:26:56 |
huggingface/transformers | ade62c2a0f4e8fbe2df9f00c733b6c04f112c135 | 558666f26f3615d3a51f3cfb0ac0a5fe5fcc24e0 | [`Tokenizers`] Change treatment of special tokens (#42903)
* fix
* test | [
{
"path": "src/transformers/tokenization_utils_base.py",
"patch": "@@ -2152,9 +2152,10 @@ def save_pretrained(\n # Add tokenizer class to the tokenizer config to be able to reload it with from_pretrained\n tokenizer_class = self.__class__.__name__\n \n- # tokenizers backend don't need... | 2025-12-17T13:08:34 |
ggml-org/llama.cpp | e4832e3ae4d58ac0ecbdbf4ae055424d6e628c9f | 960e5e3b46f211b3c4446ce8380e88404274dbce | vocab : fix attribute overrides for harmony (#18806)
* vocab : fix attribute overrides for harmony
* cont : add warning log | [
{
"path": "src/llama-vocab.cpp",
"patch": "@@ -2436,7 +2436,10 @@ void llama_vocab::impl::load(llama_model_loader & ml, const LLM_KV & kv) {\n auto & attr = id_to_token[t.second].attr;\n \n if (t.first == \"<|channel|>\" || t.first == \"<|message|>\" || t.first == \"<|start|>\" || t.... | 2026-01-13T15:40:13 |
ollama/ollama | 206797bda4685288de9a775b1536b9cbde0a7246 | 43ed358f9a894c92a24d06346ad81dc76aa52cfb | Fix concurrency integration test to work locally
This worked remotely but wound up trying to spawn multiple servers
locally which doesn't work | [
{
"path": "integration/concurrency_test.go",
"patch": "@@ -45,10 +45,18 @@ func TestMultiModelConcurrency(t *testing.T) {\n \twg.Add(len(req))\n \tctx, cancel := context.WithTimeout(context.Background(), time.Second*120)\n \tdefer cancel()\n+\n+\tclient, _, cleanup := InitServerConnection(ctx, t)\n+\tdefer ... | 2024-05-23T20:12:14 |
denoland/deno | 71ac11e15ad5ea5088458e1a0b4a8c6109a18f07 | 099320204ddaad99adf11b72a6bfe30788135719 | docs: fix duplicated wording in LSP symbol comments (#32368) | [
{
"path": "cli/lsp/tsc.rs",
"patch": "@@ -2311,7 +2311,7 @@ impl NavigateToItem {\n }\n \n // The field `deprecated` is deprecated but SymbolInformation does not have\n- // a default, therefore we have to supply the deprecated deprecated\n+ // a default, therefore we have to supply the depreca... | 2026-03-24T07:47:59 |
huggingface/transformers | 558666f26f3615d3a51f3cfb0ac0a5fe5fcc24e0 | 1dc69bd6f7f5dcc1a32467be505dd873d573846e | Preprocessing fixes and more tests for LFM2-VL (#42784)
* fix processing bugs + add more test cases
* add more image processor tests
* refactor expand_text_with_placeholders
* CI fix | [
{
"path": "src/transformers/models/lfm2_vl/processing_lfm2_vl.py",
"patch": "@@ -165,63 +165,103 @@ def expand_text_with_placeholders(\n image_sizes: list[list[int]],\n use_image_special_tokens: bool,\n **images_kwargs,\n- ):\n- prompt_strings = []\n+ ) -> list[str]:\n+ ... | 2025-12-17T12:36:42 |
vuejs/vue | 6e9fcfc81d922a1b188268bf50d7e67c07d6d662 | b7f7f2756928f409950186c5d641034f362b392a | feat(errors): sync/async error handling for lifecycle hooks and v-on handlers (#8395)
close #6953, close #7653 | [
{
"path": "src/core/instance/events.js",
"patch": "@@ -5,7 +5,8 @@ import {\n toArray,\n hyphenate,\n handleError,\n- formatComponentName\n+ formatComponentName,\n+ handlePromiseError\n } from '../util/index'\n import { updateListeners } from '../vdom/helpers/index'\n \n@@ -135,7 +136,8 @@ export f... | 2018-12-20T15:02:28 |
ggml-org/llama.cpp | 960e5e3b46f211b3c4446ce8380e88404274dbce | 20ca2e12c40bb1b17b22b8acef22bf559a6f91c3 | llama-mmap: fix direct-io loading fallback EOF exception (#18801) | [
{
"path": "src/llama-mmap.cpp",
"patch": "@@ -244,11 +244,14 @@ struct llama_file::impl {\n }\n errno = 0;\n if (fd == -1) {\n- std::size_t ret = std::fread(ptr, len, 1, fp);\n+ const size_t curr_off = tell();\n+ const size_t to_read = std::min(len, s... | 2026-01-13T14:57:07 |
ollama/ollama | fb9cdfa72335993e454af1c0eee2235f2a7e88a4 | efac488675bb4118b6dab2587e481e8640102fad | Fix server.cpp for the new cuda build macros | [
{
"path": "llm/ext_server/server.cpp",
"patch": "@@ -2335,9 +2335,9 @@ static void server_params_parse(int argc, char **argv, server_params &sparams, g\n invalid_param = true;\n break;\n }\n-#ifndef GGML_USE_CUBLAS\n- fprintf(stderr, \"warning: llama.cp... | 2024-05-18T23:02:13 |
denoland/deno | 099320204ddaad99adf11b72a6bfe30788135719 | 0558eecc0273358e962d059f1a8105d32b1d211b | fix(inspector): send executionContextDestroyed on process.exit() (#32934)
## Summary
- When `process.exit()` is called during an inspector debugging session,
broadcast `Runtime.executionContextDestroyed` CDP notification to all
connected sessions before exiting
- Wait for debugger sessions to disconnect after sending... | [
{
"path": "cli/worker.rs",
"patch": "@@ -94,6 +94,7 @@ impl CliMainWorker {\n {\n let coverage_for_exit = coverage_cell.clone();\n let profiler_for_exit = profiler_cell.clone();\n+ let inspector = self.worker.js_runtime().inspector();\n let mut cbs = OpExitCallbacks::default();\n ... | 2026-03-24T07:36:24 |
vuejs/vue | b7f7f2756928f409950186c5d641034f362b392a | 080dd971f77f7c631650c4e3027d1802f4e804d8 | feat: use event delegation when possible
This also fixes async edge case #6566 where events propagate too slow
and incorrectly trigger handlers post-patch. | [
{
"path": "src/core/util/env.js",
"patch": "@@ -14,6 +14,7 @@ export const isEdge = UA && UA.indexOf('edge/') > 0\n export const isAndroid = (UA && UA.indexOf('android') > 0) || (weexPlatform === 'android')\n export const isIOS = (UA && /iphone|ipad|ipod|ios/.test(UA)) || (weexPlatform === 'ios')\n export c... | 2018-12-20T00:26:58 |
huggingface/transformers | 0dbf808573cb3433992a9d4177222fdccd67b4ee | a33ef4f94eb0df76147fa1eff69a39e0f875b8f7 | adds jais2 model support (#42684)
* adds jais2 model support
* updates tests
* addresses review comment
* review comments addressed
* addresses test review comments
* fixes date
* format issue fix
* Update src/transformers/models/jais2/__init__.py
Co-authored-by: Anton Vlasjuk <73884904+vasqu@users.noreply.git... | [
{
"path": "docs/source/en/_toctree.yml",
"patch": "@@ -551,6 +551,8 @@\n title: HunYuanMoEV1\n - local: model_doc/ibert\n title: I-BERT\n+ - local: model_doc/jais2\n+ title: Jais2\n - local: model_doc/jamba\n title: Jamba\n - local: model_doc/jetmoe",
... | 2025-12-17T11:06:04 |
ggml-org/llama.cpp | ea4a321f2a607ca315d998f0656fd255715884a6 | c1e79e610fd28f2c3923539fee9313734bbf8cfa | HIP: add fattn-mma-f16 for RDNA4 (#18481)
* finish VQ mma
* flash_attn_ext_f16_iter
* KQ_rowsum
* correct exp
* fix scale error
* fix softmax scale
* fix softmax scale
* enable fattn on cpu side
* fix random error
* disable fattn-mma-f16 on rdna3
* fix wrong col for rdna
* use identity mat to transpose
* r... | [
{
"path": "ggml/src/ggml-cuda/common.cuh",
"patch": "@@ -262,6 +262,10 @@ static const char * cu_get_error_str(CUresult err) {\n #define FLASH_ATTN_AVAILABLE\n #endif // !defined(GGML_CUDA_NO_FA) && !(defined(GGML_USE_MUSA) && __MUSA_ARCH__ < 220)\n \n+#if defined(TURING_MMA_AVAILABLE)\n+#define LDMATRIX_TR... | 2026-01-13T12:52:16 |
vuejs/vue | 080dd971f77f7c631650c4e3027d1802f4e804d8 | 850555d1faa9be7d8306adffd95c7dee5e58717f | fix(next-tick): revert 60da366
the condition is no longer necessary after reverting back to microtask
only nextTick implementation, and fix #8436 | [
{
"path": "src/platforms/web/runtime/modules/dom-props.js",
"patch": "@@ -86,10 +86,6 @@ function isDirtyWithModifiers (elm: any, newVal: string): boolean {\n const value = elm.value\n const modifiers = elm._vModifiers // injected by v-model runtime\n if (isDef(modifiers)) {\n- if (modifiers.lazy) ... | 2018-12-19T19:00:15 |
denoland/deno | f54ea9f126b4f73bc5b6af8622b43340a074bb7c | e7eb863c527b7441521d567a48b3f9af3f56ef34 | chore: consolidate Copilot review instructions and add runtime-aware guidelines (#32938)
## Summary
- Merge `.github/instructions/code-review.instructions.md` into
`.github/copilot-instructions.md`, eliminating the separate instructions
directory
- Add new review guidelines based on recurring patterns of incorrect
Co... | [
{
"path": ".github/copilot-instructions.md",
"patch": "@@ -228,6 +228,123 @@ RUST_BACKTRACE=1 ./target/debug/deno run script.ts\n \n In Rust code: `eprintln!(\"Debug: {:?}\", var);` or `dbg!(var);`\n \n+## Pull Request Reviews\n+\n+### Before commenting, verify your claims\n+\n+- If you claim something is m... | 2026-03-24T07:10:31 |
huggingface/transformers | a33ef4f94eb0df76147fa1eff69a39e0f875b8f7 | dd24a80666b72c85f02c6cf9df18164cc174ab74 | Rewrite for loop in get_image_features with torch ops for export (#42822)
* Rewrite ministral 3 for loop with torch ops for export
* Fix Llava as well
* Format
* make fix-copies
* Accidentally reversed ministral changes
* Update modular mistral3
* Update .md
* Update .md | [
{
"path": "src/transformers/models/llava/modeling_llava.py",
"patch": "@@ -202,10 +202,11 @@ def get_image_features(\n image_features = self.multi_modal_projector(selected_image_feature)\n \n if \"image_sizes\" in kwargs:\n- split_sizes = [\n- (height // self.vision... | 2025-12-17T09:53:37 |
ggml-org/llama.cpp | e047f9ee9d6d54c5c7d58de82c71103ca2c26b6c | 0a57271ab66ac52057e13f6658ecedf5d338c0be | mtmd: fix use_non_causal being reported incorrectly (#18793)
* mtmd: fix use_non_causal being reported incorrectly
* move clip_is_mrope to mtmd_decode_use_mrope
* fix sloppy code ggml_cpy | [
{
"path": "src/models/gemma3n-iswa.cpp",
"patch": "@@ -258,12 +258,12 @@ ggml_tensor * llm_build_gemma3n_iswa::get_per_layer_inputs() {\n res->add_input(std::move(inp));\n } else {\n // Vision embedding path: use padding token (ID=0) embedding\n+ // TODO: verify if this is the cor... | 2026-01-13T11:19:38 |
ollama/ollama | d528e1af75e5b253b21ec90fa9d5f33737fd4909 | cd234ce22c85bf34dc50b05c93c4dab513ae8f99 | fix utf16 for multibyte runes | [
{
"path": "parser/parser.go",
"patch": "@@ -8,7 +8,9 @@ import (\n \t\"io\"\n \t\"strconv\"\n \t\"strings\"\n-\t\"unicode\"\n+\n+\t\"golang.org/x/text/encoding/unicode\"\n+\t\"golang.org/x/text/transform\"\n )\n \n type File struct {\n@@ -69,14 +71,11 @@ func ParseFile(r io.Reader) (*File, error) {\n \tvar ... | 2024-06-13T18:39:01 |
vuejs/vue | 850555d1faa9be7d8306adffd95c7dee5e58717f | ced774be6ddbc53884d7a5d395514a9f62e32336 | fix: always use microtasks for nextTick (#8450)
fix #7109, #7546, #7707, #7834, #8109
reopen #6566 | [
{
"path": "src/core/util/next-tick.js",
"patch": "@@ -1,9 +1,9 @@\n /* @flow */\n-/* globals MessageChannel */\n+/* globals MutationObserver */\n \n import { noop } from 'shared/util'\n import { handleError } from './error'\n-import { isIOS, isNative } from './env'\n+import { isIE, isIOS, isNative } from '.... | 2018-12-19T18:29:00 |
denoland/deno | e7eb863c527b7441521d567a48b3f9af3f56ef34 | 7a2a390432fc3eddef21db0ea6db6c392f63d13d | fix(npm): skip `file:` and `link:` dependencies in npm packages (#32876)
## Summary
- npm packages with `file:` or `link:` dependency specifiers (e.g.
`"local-pkg": "file:./local-pkg"`) caused Deno to fail with "Invalid
version requirement. Unexpected character." during resolution
- These specifiers are used for local... | [
{
"path": "libs/npm/registry.rs",
"patch": "@@ -131,6 +131,11 @@ pub enum NpmDependencyEntryErrorSource {\n \n To work around this, you can use a package.json and install the dependencies via `npm install`.\", .specifier)]\n RemoteDependency { specifier: String },\n+ /// A `file:` or `link:` dependency t... | 2026-03-23T20:05:17 |
ggml-org/llama.cpp | 0a57271ab66ac52057e13f6658ecedf5d338c0be | 076b0faf7ddbcf9c1c68b90be2a47497a57141c2 | CUDA : fix unused argument when USE_CUDA_GRAPH=OFF (#18800) | [
{
"path": "ggml/src/ggml-cuda/ggml-cuda.cu",
"patch": "@@ -3737,6 +3737,7 @@ static bool ggml_cuda_graph_set_enabled(ggml_backend_cuda_context * cuda_ctx) {\n \n return cuda_ctx->cuda_graph->is_enabled();\n #else\n+ GGML_UNUSED(cuda_ctx);\n return false;\n #endif // USE_CUDA_GRAPH\n }",
"addi... | 2026-01-13T10:25:53 |
huggingface/transformers | dd24a80666b72c85f02c6cf9df18164cc174ab74 | 703da867001c5e2c8322b2c3395e6c4a5ba31022 | Support having multiple sub-processors (of any kind) in the same processor (#42667)
* support saving/loading multiple sub_processor of the same kind
* standardize all processors
* remove tokenizer_class from lasr
* fix modular
* fix kwargs logic
* override _load_tokenizer_from_pretrained in pixtral and fuyu
* fi... | [
{
"path": "src/transformers/models/audioflamingo3/processing_audioflamingo3.py",
"patch": "@@ -74,10 +74,6 @@ class AudioFlamingo3Processor(ProcessorMixin):\n Special token used to represent audio inputs in the chat template.\n \"\"\"\n \n- attributes = [\"feature_extractor\", \"tokenizer... | 2025-12-16T22:11:54 |
denoland/deno | 70268d6b9d02a6fb46b75ea444a3b9bd7779469c | dcc04bbd62563016a30a8fa4eb66f4598cae857b | feat(core): implement NodeRuntime CDP domain and fix --inspect-brk blocking (#32714)
## Summary
Implements the `NodeRuntime` CDP (Chrome DevTools Protocol) domain for
the inspector, and fixes several long-standing inspector issues.
### NodeRuntime CDP Domain
- `NodeRuntime.enable` — emits `NodeRuntime.waitingForDebu... | [
{
"path": "libs/core/inspector.rs",
"patch": "@@ -194,7 +194,9 @@ impl v8::inspector::V8InspectorClientImpl for JsRuntimeInspectorClient {\n \n fn run_if_waiting_for_debugger(&self, context_group_id: i32) {\n assert_eq!(context_group_id, JsRuntimeInspector::CONTEXT_GROUP_ID);\n- self.0.flags.borrow... | 2026-03-23T13:08:12 |
ollama/ollama | 1fd236d1776445b45af104d259a0d997ef0b1806 | e87fc7200d47bd2e5214126802da1d5607b686be | server: remove jwt decoding error (#5027) | [
{
"path": "server/images.go",
"patch": "@@ -960,7 +960,6 @@ var errUnauthorized = fmt.Errorf(\"unauthorized: access denied\")\n func getTokenSubject(token string) string {\n \tparts := strings.Split(token, \".\")\n \tif len(parts) != 3 {\n-\t\tslog.Error(\"jwt token does not contain 3 parts\")\n \t\treturn ... | 2024-06-13T18:21:15 |
vuejs/vue | b06c784b81a244e1bc2d028216fcd2ab873730b9 | 70754084ec2d84e44effeb8f82efda3397a08b0d | fix(ssr): support rendering comment (#9128) | [
{
"path": "src/server/optimizing-compiler/codegen.js",
"patch": "@@ -225,7 +225,11 @@ function nodesToSegments (\n } else if (c.type === 2) {\n segments.push({ type: INTERPOLATION, value: c.expression })\n } else if (c.type === 3) {\n- segments.push({ type: RAW, value: escape(c.text) })\n... | 2018-12-12T16:41:08 |
ggml-org/llama.cpp | 076b0faf7ddbcf9c1c68b90be2a47497a57141c2 | db79dc06b14d8d4c5f6b49c0191444689e6a5db3 | graph : clean up t5 input builders (#18795)
* fix: Remove unnecessary `h` loops where `h` was only ever 0
Branch: CleanUpT5InputBuilders
Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>
* fix: Remove unnecessary padding loop that is never hit anymore
The upper bound used to use GGML_PAD(n_tokens, GGML_KQ_MASK_PAD),... | [
{
"path": "src/llama-graph.cpp",
"patch": "@@ -96,11 +96,9 @@ void llm_graph_input_pos_bucket::set_input(const llama_ubatch * ubatch) {\n \n int32_t * data = (int32_t *) pos_bucket->data;\n \n- for (int h = 0; h < 1; ++h) {\n- for (int j = 0; j < n_tokens; ++j) {\n- ... | 2026-01-13T08:43:51 |
huggingface/transformers | 703da867001c5e2c8322b2c3395e6c4a5ba31022 | 9aef5ca4d7d3e1a651c8a06071bda4aac094363c | Fix add_dates script: Fetch github repo from url to check if model is new (#42878)
Fetch github repo from url to check if model is new | [
{
"path": "docs/source/en/model_doc/paddleocr_vl.md",
"patch": "@@ -13,7 +13,7 @@ specific language governing permissions and limitations under the License.\n rendered properly in your Markdown viewer.\n \n -->\n-*This model was released on 2025.10.16 and added to Hugging Face Transformers on 2025.12.10*\n+... | 2025-12-16T18:30:42 |
vuejs/vue | 847e493768371cec4718969e02bdb7f8463f4e03 | 4e975485a4d90fef62bf73a32a2b72e249919130 | fix: fix single v-for child optimization | [
{
"path": "src/core/instance/render-helpers/render-list.js",
"patch": "@@ -32,8 +32,9 @@ export function renderList (\n ret[i] = render(val[key], key, i)\n }\n }\n- if (isDef(ret)) {\n- (ret: any)._isVList = true\n+ if (!isDef(ret)) {\n+ ret = []\n }\n+ (ret: any)._isVList = true\n ... | 2018-12-11T22:47:59 |
denoland/deno | dcc04bbd62563016a30a8fa4eb66f4598cae857b | b75e2b8dda4c495d56dfc8126893c1271c012199 | fix(ext/node): defer uv_write callbacks to prevent re-entrancy panic (#32919)
## Summary
- `uv_write` in the uv compat layer was firing write callbacks
synchronously when `try_write` succeeded fully or failed. This caused a
`RefCell` double-borrow panic ("RefCell already borrowed") when the
caller (e.g. `StreamWrap.w... | [
{
"path": "libs/core/uv_compat/stream.rs",
"patch": "@@ -254,10 +254,14 @@ pub unsafe fn uv_write(\n data: write_data,\n offset: 0,\n cb,\n+ status: None,\n });\n return 0;\n }\n \n+ // Never fire callbacks synchronously from uv_write — always queue.\n+ /... | 2026-03-23T11:19:38 |
ggml-org/llama.cpp | 4150da9a956d0587a3e8a08e69940f9a27f88e5c | 8e2da778dab89990590c68c69db06208a9c11e31 | examples : add --kv-unified to batched example (#18774)
This commit adds the --kv-unified flag to the batched example. This flag
is currently specified in the README.md as required, but is currently
not available as a command line option for the batched example.
The motivation for this is that specifying this flag as... | [
{
"path": "common/arg.cpp",
"patch": "@@ -1295,7 +1295,7 @@ common_params_context common_params_parser_init(common_params & params, llama_ex\n [](common_params & params) {\n params.kv_unified = true;\n }\n- ).set_env(\"LLAMA_ARG_KV_UNIFIED\").set_examples({LLAMA_EXAMPLE_SERVER... | 2026-01-12T12:47:58 |
vuejs/vue | 47487607fbb99339038cf84990ba341c25b5e20d | 984393fed981c58ad79ed50424f023dcfa6829d0 | fix: fix v-for component with undefined value
fix #9181 | [
{
"path": "src/core/vdom/helpers/normalize-children.js",
"patch": "@@ -16,6 +16,9 @@ import { isFalse, isTrue, isDef, isUndef, isPrimitive } from 'shared/util'\n // thing with Array.prototype.concat. It is guaranteed to be only 1-level deep\n // because functional components already normalize their own chil... | 2018-12-11T21:51:03 |
denoland/deno | b75e2b8dda4c495d56dfc8126893c1271c012199 | d0e3cfb1ae8c87cd3108bd7870eba590de6377f3 | fix(ext/node): fix node:domain across async boundaries (#32897)
## Summary
- Fix `node:domain` not working across async boundaries (timers, I/O).
`process.domain` was lost in timer callbacks and socket error handlers
because
timers didn't emit async hook lifecycle events and `EventEmitter.emit`
didn't
preserve doma... | [
{
"path": "ext/node/polyfills/domain.ts",
"patch": "@@ -121,9 +121,28 @@ export class Domain extends EventEmitter {\n // deno-lint-ignore no-this-alias\n const self = this;\n return function () {\n+ self.enter();\n try {\n- return FunctionPrototypeApply(fn, null, ArrayPrototype... | 2026-03-23T09:40:44 |
ollama/ollama | c16f8af91150febf705cefe28d48c90979cece30 | 217f60c3d95edc7d5dcbeb4c3cffb0190c147f92 | fix: multiple templates when creating from model
multiple templates may appear in a model if a model is created from
another model that 1) has an autodetected template and 2) defines a
custom template | [
{
"path": "server/images.go",
"patch": "@@ -28,7 +28,6 @@ import (\n \t\"github.com/ollama/ollama/format\"\n \t\"github.com/ollama/ollama/llm\"\n \t\"github.com/ollama/ollama/parser\"\n-\t\"github.com/ollama/ollama/templates\"\n \t\"github.com/ollama/ollama/types/errtypes\"\n \t\"github.com/ollama/ollama/ty... | 2024-06-12T20:30:08 |
huggingface/transformers | 9aef5ca4d7d3e1a651c8a06071bda4aac094363c | b1a2fba1c47e14390f17998218addf20c329e010 | Sam: Perception Encoder Audiovisual (#42905)
* Adding PE-Audio model
* Adding judge and FL-CLAP models
* lint
* Update to flatten ModernBERT
* Remove EMPTY_DICT default parameters
* Use AutoModel for modernbert
* Reference config directly instead of adding new properties to model
* Remove patch_size from `Patch... | [
{
"path": "docs/source/en/_toctree.yml",
"patch": "@@ -921,6 +921,8 @@\n title: MusicGen Melody\n - local: model_doc/parakeet\n title: Parakeet\n+ - local: model_doc/pe_audio\n+ title: PE Audio\n - local: model_doc/pop2piano\n title: Pop2Piano\n - local:... | 2025-12-16T17:54:30 |
ggml-org/llama.cpp | 2bbe4c2cf8298114e3908e285125b9d0d1c5bb42 | 1051ecd28907d2ca0a15c135f190fe415d0a3d1b | vulkan: Use VK_EXT_shader_64bit_indexing to handle large mat_mul(_id) (#18678)
This fixes incoherent output in Llama-4-Maverick-17B-128E-PAB-Q8_0, which
has a mul_mat_id with an A matrix that's Q8_0 8192 x 5120 x 128.
This should work when the number of blocks in the A matrix is less than 2^32
(for mul_mat_vec or mul... | [
{
"path": "ggml/src/ggml-vulkan/ggml-vulkan.cpp",
"patch": "@@ -119,6 +119,8 @@ struct ggml_backend_vk_context;\n // Max number of adds that can be fused without exceeding MAX_PARAMETER_COUNT.\n #define MAX_FUSED_ADDS (MAX_PARAMETER_COUNT - 3)\n \n+typedef std::shared_ptr<struct vk_pipeline_struct> vk_pipel... | 2026-01-12T11:32:13 |
vuejs/vue | 87bad80f0cb9a30b95d9410120ff6e3e2022a723 | d8285c57a613c42eddf2d4f2b75c1cea6aa4703a | fix(lifecycle): beforeUpdated should not be called if component is destroyed (#9171)
fix #8076 | [
{
"path": "src/core/instance/lifecycle.js",
"patch": "@@ -196,7 +196,7 @@ export function mountComponent (\n // component's mounted hook), which relies on vm._watcher being already defined\n new Watcher(vm, updateComponent, noop, {\n before () {\n- if (vm._isMounted) {\n+ if (vm._isMounted... | 2018-12-11T16:31:27 |
denoland/deno | d0e3cfb1ae8c87cd3108bd7870eba590de6377f3 | e8ba184376aaf1463ba915fa8ac27ad558ddc868 | fix(ext/web): AbortSignal.any() memory leak with long-lived signals (#32916)
## Summary
`AbortSignal.any()` leaked memory when called repeatedly with a
long-lived parent signal (common pattern: a global shutdown signal
combined per-request). The parent signal's `dependentSignals` collection
grew unboundedly because d... | [
{
"path": "ext/web/03_abort_signal.js",
"patch": "@@ -5,22 +5,19 @@\n \n import { core, primordials } from \"ext:core/mod.js\";\n const {\n- ArrayPrototypeEvery,\n ArrayPrototypePush,\n FunctionPrototypeApply,\n ObjectPrototypeIsPrototypeOf,\n+ SafeFinalizationRegistry,\n SafeSet,\n SafeSetItera... | 2026-03-23T09:37:58 |
huggingface/transformers | b1a2fba1c47e14390f17998218addf20c329e010 | 6994c5ac32d0f930030a330d396b8e8d248161a8 | fix `Dtensor` and `tensor` mismatch (#42906)
* begin Moe test tensor parallel
* create tiny moe model + fix test tensor parallel Moe
eaeaae
* create tiny moe model + fix test tensor parallel Moe
eaeaae
fix tensor parallel MoE test
fix tensor parallel MoE test
* fix backward pass test in tensor parallel for Dense... | [
{
"path": "src/transformers/integrations/tensor_parallel.py",
"patch": "@@ -758,6 +758,15 @@ def partition_tensor(self, param, empty_param, param_type, param_casting_dtype,\n return nn.Parameter(parameter, requires_grad=parameter.is_floating_point())\n \n \n+class LocalColwiseParallel(ColwiseParalle... | 2025-12-16T17:36:04 |
vuejs/vue | d8285c57a613c42eddf2d4f2b75c1cea6aa4703a | 8fccd8256902b6199ac124017f70d75abb45ed27 | fix(weex): support data class type that is string (#9139)
close #9124 | [
{
"path": "src/platforms/weex/runtime/modules/class.js",
"patch": "@@ -44,6 +44,8 @@ function makeClassList (data: VNodeData): Array<string> {\n classList.push.apply(classList, dataClass)\n } else if (isObject(dataClass)) {\n classList.push.apply(classList, Object.keys(dataClass).filter(className ... | 2018-12-11T16:29:49 |
denoland/deno | e8ba184376aaf1463ba915fa8ac27ad558ddc868 | 80121ae106e863157920e7709b5a08e67505195f | fix(ext/node): fix IPC stdio handling in child_process (#32920)
## Summary
Two fixes for child_process IPC stdio handling:
1. **Validate single IPC pipe** - throw `ERR_IPC_ONE_PIPE` when multiple
`'ipc'` entries are specified in the stdio array (e.g. `stdio: ['pipe',
'pipe', 'pipe', 'ipc', 'ipc']`)
2. **Fix panic w... | [
{
"path": "ext/node/polyfills/internal/child_process.ts",
"patch": "@@ -46,6 +46,7 @@ import {\n ERR_INVALID_HANDLE_TYPE,\n ERR_INVALID_SYNC_FORK_INPUT,\n ERR_IPC_CHANNEL_CLOSED,\n+ ERR_IPC_ONE_PIPE,\n ERR_IPC_SYNC_FORK,\n ERR_MISSING_ARGS,\n ERR_UNKNOWN_SIGNAL,\n@@ -382,6 +383,9 @@ export clas... | 2026-03-23T09:37:00 |
ollama/ollama | 7bdcd1da9417ecc1a1069f2d73f0b88a3a43857d | 0f3cf1d42e47f5293c14422032c324a894e3dd93 | Revert "Merge pull request #4938 from ollama/mxyng/fix-byte-order"
This reverts commit f5f245cc154580fa7b4052c001d2a7e3d771cfb8, reversing
changes made to 94d37fdcae30ddeb6c9f65c8707004f5ec9eaf33.
this change broke gguf v2 which is incorrectly detected as big endian | [
{
"path": "llm/ggml.go",
"patch": "@@ -231,7 +231,8 @@ const (\n \t// Magic constant for `ggla` files (LoRA adapter).\n \tFILE_MAGIC_GGLA = 0x67676C61\n \t// Magic constant for `gguf` files (versioned, gguf)\n-\tFILE_MAGIC_GGUF = 0x46554747\n+\tFILE_MAGIC_GGUF_LE = 0x46554747\n+\tFILE_MAGIC_GGUF_BE = 0x4747... | 2024-06-11T22:55:44 |
ggml-org/llama.cpp | 0c3b7a9efebc73d206421c99b7eb6b6716231322 | 0e76501e1d2450e486481b640bb3a942977cfd3a | model: fix qwen3next broken due to #18683 (#18762) | [
{
"path": "src/models/qwen3next.cpp",
"patch": "@@ -590,6 +590,7 @@ std::pair<ggml_tensor *, ggml_tensor *> llm_build_qwen3next::build_qkvz(\n ggml_tensor * z = ggml_view_4d(ctx0, mixed_qkvz_reshaped, split_sizes_qkvz[3], num_k_heads, n_seq_tokens, n_seqs,\n mixed... | 2026-01-11T20:00:10 |
huggingface/transformers | 4c64a8fb15de548ea3a684f6a621a9c3118318ef | b61da251692e5a107117618ca71d0b1ccd2e059c | fix: typehits for Causal LM models (#42885)
chore: fix missing typecheck imports | [
{
"path": "src/transformers/models/__init__.py",
"patch": "@@ -23,6 +23,7 @@\n from .albert import *\n from .align import *\n from .altclip import *\n+ from .apertus import *\n from .arcee import *\n from .aria import *\n from .audio_spectrogram_transformer import *\n@@ -107,6 +10... | 2025-12-16T14:35:24 |
vuejs/vue | 8fccd8256902b6199ac124017f70d75abb45ed27 | 3d449376d557c4533a9664f95df3a168ecee9bfa | chore: fix tag name in compiler/parser/index.js (#9165) | [
{
"path": "src/compiler/parser/index.js",
"patch": "@@ -340,7 +340,7 @@ function processKey (el) {\n const parent = el.parent\n if (iterator && iterator === exp && parent && parent.tag === 'transition-group') {\n warn(\n- `Do not use v-for index as key on <transtion-grou... | 2018-12-11T16:28:40 |
ggml-org/llama.cpp | 9789e28459a231bed38859edbdcb10bd058c66d4 | 84ae04f163140f24eb7d1fae9c1893edbec5ca05 | debug : include LLAMA_POOLING_TYPE_UNSPECIFIED in pooling check (#18692)
* debug : include LLAMA_POOLING_TYPE_UNSPECIFIED in pooling check
This commit updates the pooling check in the debug example to
also include LLAMA_POOLING_TYPE_UNSPECIFIED and not just
LLAMA_POOLING_TYPE_NONE.
* debug : normalize both pooled an... | [
{
"path": "examples/debug/debug.cpp",
"patch": "@@ -57,11 +57,21 @@ struct callback_data {\n }\n };\n \n+static bool has_pooling(llama_context * ctx) {\n+ switch (llama_pooling_type(ctx)) {\n+ case LLAMA_POOLING_TYPE_NONE:\n+ case LLAMA_POOLING_TYPE_UNSPECIFIED:\n+ return fal... | 2026-01-11T15:34:41 |
denoland/deno | c99c8aceff594d5914351f5f0272e6291a65dff0 | 1df0225552d39cc2e04286eeb236391d59c6b8b0 | fix(ext/node): fix Buffer.concat truncating buffers larger than 4GB (#32913)
## Summary
- Replace `| 0` (32-bit integer truncation) with `MathTrunc()` in
`checked()`, `_allocUnsafe()`, and `fromString()` in the Node.js Buffer
polyfill
- This fixes `Buffer.concat` and `Buffer.allocUnsafe` silently
truncating buffers w... | [
{
"path": "ext/node/polyfills/internal/buffer.mjs",
"patch": "@@ -423,7 +423,7 @@ Buffer.alloc = function alloc(size, fill, encoding) {\n \n function _allocUnsafe(size) {\n assertSize(size);\n- return createBuffer(size < 0 ? 0 : checked(size) | 0);\n+ return createBuffer(size < 0 ? 0 : checked(size));\n... | 2026-03-23T07:38:49 |
huggingface/transformers | 24275124c69477762e1dd6af2bb05d9513098a7e | 4d6516e2565af683f6edad4ae8d13caa6a70bf0b | Add local kernel loading support to KernelConfig(). (#42800)
* add add_to_mapping_local for KernelConfig
* refactor kernel_mapping format
* lint code
* specify the kernel path
* fix `abs/path`
Co-authored-by: Mohamed Mekkouri <93391238+MekkCyber@users.noreply.github.com>
* rename check_kernel_from_local to use_l... | [
{
"path": "src/transformers/modeling_utils.py",
"patch": "@@ -3533,7 +3533,9 @@ def set_use_kernels(self, use_kernels, kernel_config):\n \n # This is a context manager to override the default kernel mapping\n # We are calling kernelize inside this context manager using the us... | 2025-12-16T13:02:30 |
ollama/ollama | ead259d877fc8b20f7943f1f9e8eeaae0acfa52a | 2ff45d571de4463fcebf779373ae7337cf969ebf | llm: fix seed value not being applied to requests (#4986) | [
{
"path": "docs/api.md",
"patch": "@@ -250,7 +250,7 @@ curl http://localhost:11434/api/generate -d '{\n \n #### Request (Reproducible outputs)\n \n-For reproducible outputs, set `temperature` to 0 and `seed` to a number:\n+For reproducible outputs, set `seed` to a number:\n \n ##### Request\n \n@@ -259,8 +2... | 2024-06-11T21:24:41 |
ggml-org/llama.cpp | 506bb6e01009058f35558474cf987eeb56361782 | 79456a690ae35cb2a75faf24d3b1926f716b0485 | model: try to improve Qwen3 Next (#18683)
* qwen3next: simplify qkvz projection
* use ggml_swiglu_split
* revert swiglu_split, but remove redundant repeat()
* fix missing reshape
* rm 2 redundant transposes
* move mul_mat(k,q) to outside of chunking
* rm redundant cont
* improve g_cs_chunk
* add comments about... | [
{
"path": "convert_hf_to_gguf.py",
"patch": "@@ -4367,7 +4367,37 @@ def modify_tensors(self, data_torch: Tensor, name: str, bid: int | None) -> Iter\n elif name.endswith(\"norm.weight\") and not name.endswith(\"linear_attn.norm.weight\"):\n data_torch = data_torch + 1\n \n- yield ... | 2026-01-11T11:53:33 |
huggingface/transformers | 4d6516e2565af683f6edad4ae8d13caa6a70bf0b | 24b311eead0ee199ed6e77f8b562f01b7afa924d | Simplify tie weights logic (#42895)
* fix
* let's not use source backup, clearer to use original name imo
* fix
* use a set
* simplify
* style
* add comment | [
{
"path": "src/transformers/modeling_utils.py",
"patch": "@@ -2387,11 +2387,6 @@ def tie_weights(self, missing_keys: Optional[set[str]] = None, recompute_mapping\n \n tied_keys = list(tied_keys.items())\n for i, (target_param_name, source_param_name) in enumerate(tied_keys):\n- # ... | 2025-12-16T11:58:21 |
denoland/deno | 1df0225552d39cc2e04286eeb236391d59c6b8b0 | 72d86f79649efc4a9777235415af82accfa02a03 | fix(ext/node): set fd property on TTY ReadStream and WriteStream (#32880)
## Summary
- Set `this.fd = fd` in tty `WriteStream` and `ReadStream` constructors
- When stdout/stderr is a TTY, `process.stdout.fd` was `undefined`
because the TTY code path in `process.ts` creates `TTYWriteStream`
directly, bypassing `create... | [
{
"path": "ext/node/polyfills/process.ts",
"patch": "@@ -1317,6 +1317,9 @@ internals.__bootstrapNodeProcess = function (\n if (io.stdout.isTerminal()) {\n /** https://nodejs.org/api/process.html#process_process_stdout */\n stdout = process.stdout = new TTYWriteStream(1);\n+ // For suppo... | 2026-03-23T07:35:32 |
vuejs/vue | 3d449376d557c4533a9664f95df3a168ecee9bfa | d780dd2e2adcf71f40c086055a659a9a2b4a8282 | fix(v-model): properly handle multiline v-model expressions (#9184)
fix #9183 | [
{
"path": "src/compiler/directives/model.js",
"patch": "@@ -25,7 +25,7 @@ export function genComponentModel (\n \n el.model = {\n value: `(${value})`,\n- expression: `\"${value}\"`,\n+ expression: JSON.stringify(value),\n callback: `function (${baseValueExpression}) {${assignment}}`\n }\n ... | 2018-12-11T15:27:43 |
ollama/ollama | 157f09acdfd692230075a67783510b8d2bb0d5b0 | 0f3cf1d42e47f5293c14422032c324a894e3dd93 | fix: "Skip searching for network devices"
On an Ubuntu 24.04 computer with vmware installed, the sudo lshw command will get stuck. "Network interfaces" is always displayed | [
{
"path": "scripts/install.sh",
"patch": "@@ -159,8 +159,8 @@ check_gpu() {\n esac ;;\n lshw)\n case $2 in\n- nvidia) available lshw && $SUDO lshw -c display -numeric | grep -q 'vendor: .* \\[10DE\\]' || return 1 ;;\n- amdgpu) available lshw && $... | 2024-06-11T08:11:35 |
ggml-org/llama.cpp | 28068af78996485eaa5c34557eae56dc1f996f92 | 707cbafcaa3a88d893147bcc107e76b925a867da | security: narrow down the scope of what we consider a vulnerability (#18752)
* security: narrow down the scope of what we consider a vulnerability
* fix typo | [
{
"path": "SECURITY.md",
"patch": "@@ -1,12 +1,48 @@\n # Security Policy\n \n+ - [**Reporting a vulnerability**](#reporting-a-vulnerability)\n+ - [**Requirements**](#requirements)\n+ - [**Covered Topics**](#covered-topics)\n - [**Using llama.cpp securely**](#using-llamacpp-securely)\n - [Untrusted model... | 2026-01-11T11:23:36 |
huggingface/transformers | 24b311eead0ee199ed6e77f8b562f01b7afa924d | 0f896619724ad8bfe43bae91d0208f80817e5a5b | fix FastSpeech2ConformerTokenizer crash in tokenize (#42888)
Signed-off-by: Wang, Yi <yi.a.wang@intel.com> | [
{
"path": "src/transformers/models/fastspeech2_conformer/tokenization_fastspeech2_conformer.py",
"patch": "@@ -79,6 +79,7 @@ def __init__(\n unk_token=unk_token,\n pad_token=pad_token,\n should_strip_spaces=should_strip_spaces,\n+ special_tokens_pattern=\"none\... | 2025-12-16T11:42:38 |
denoland/deno | 72d86f79649efc4a9777235415af82accfa02a03 | af670585e4d144b8c9a5863f1b44d68a58544f8b | fix(ext/node): enable node:fs compat tests + lchmod fix (#32892)
## Summary
Enable 87 previously-disabled `node:fs` compatibility tests that now
pass.
### Code fix
- **promises.lchmod**: Made the function `async` so validation errors
become rejected promises instead of synchronous throws, matching Node.js
behavior w... | [
{
"path": "ext/node/polyfills/_fs/_fs_fstat.ts",
"patch": "@@ -12,6 +12,7 @@ import {\n import { BigIntStats, Stats } from \"ext:deno_node/internal/fs/utils.mjs\";\n import { FsFile } from \"ext:deno_fs/30_fs.js\";\n import { denoErrorToNodeError } from \"ext:deno_node/internal/errors.ts\";\n+import { getVa... | 2026-03-23T07:22:11 |
vuejs/vue | d780dd2e2adcf71f40c086055a659a9a2b4a8282 | 77d406b2a49b286fe6587305fabcf0796db19ab1 | fix(types): accept primatives and falsy values in createElement children (#9154)
fixes #8498 | [
{
"path": "types/test/options-test.ts",
"patch": "@@ -281,6 +281,12 @@ Vue.component('provide-function', {\n })\n })\n \n+Vue.component('component-with-slot', {\n+ render (h): VNode {\n+ return h('div', this.$slots.default)\n+ }\n+})\n+\n Vue.component('component-with-scoped-slot', {\n render (h) {... | 2018-12-10T23:36:41 |
ggml-org/llama.cpp | b1377188784f9aea26b8abde56d4aee8c733eec7 | d2ff4e23acd0724b44e0af72fd7e37fed4c1a6a6 | test-backend-ops: fix mxfp4 tests on blackwell (#18736) | [
{
"path": "tests/test-backend-ops.cpp",
"patch": "@@ -454,6 +454,28 @@ static bool ggml_is_view_op(enum ggml_op op) {\n return op == GGML_OP_VIEW || op == GGML_OP_RESHAPE || op == GGML_OP_PERMUTE || op == GGML_OP_TRANSPOSE;\n }\n \n+static bool backend_has_feature(ggml_backend_t backend, const char * fe... | 2026-01-10T17:12:57 |
huggingface/transformers | 0f896619724ad8bfe43bae91d0208f80817e5a5b | 5d2f82b530d9e53fdb64b4e78638cd8f1673378f | Added kernels from kernel hub for Bamba model (#41540)
* Added kernels from kernel hub for Bamba model
* Updated kernel loading
Signed-off-by: romit <romit@ibm.com>
* Remove einops
Signed-off-by: romit <romit@ibm.com>
* Removed global vars
Signed-off-by: romit <romit@ibm.com>
* Fixed make style
Signed-off-by: ... | [
{
"path": "src/transformers/integrations/hub_kernels.py",
"patch": "@@ -370,7 +370,7 @@ def lazy_load_kernel(kernel_name: str, mapping: dict[str, ModuleType | None] = _\n if callable(is_kernel_available) and is_kernel_available():\n # Try to import the module \"{kernel_name}\" from paren... | 2025-12-16T11:33:49 |
vuejs/vue | e06d2af276fc8d626a3b048f4d138a243aa690a4 | 628c1b7f5b298e13975d880c73e8fc2893628c2e | fix(types): avoid `this` in VueConstructor signature (#9173)
https://github.com/vuejs/vue-class-component/issues/294#issuecomment-445526936 | [
{
"path": "types/test/tsconfig.json",
"patch": "@@ -1,6 +1,7 @@\n {\n \"compilerOptions\": {\n \"target\": \"es5\",\n+ \"experimentalDecorators\": true,\n \"lib\": [\n \"dom\",\n \"es2015\"",
"additions": 1,
"deletions": 0,
"language": "JSON"
},
{
"path": "types/... | 2018-12-10T16:53:57 |
denoland/deno | af670585e4d144b8c9a5863f1b44d68a58544f8b | ceb6936a983904511867d69986203786b68b7e21 | fix(ext/node): preserve TLS upgrade state on reconnect (#32838)
## Summary
- run the TLS post-connect upgrade hook against the active handle
instead of a closure-captured original handle
- rebuild TLS upgrade promise/resolver metadata when `net.Socket`
reinitializes its handle during reconnect fallback
- add a regress... | [
{
"path": "ext/node/ops/http2/session.rs",
"patch": "@@ -20,14 +20,9 @@ use deno_core::v8;\n use libnghttp2 as ffi;\n use serde::Serialize;\n \n-/// Type alias matching the C `ssize_t` type used by nghttp2 callbacks.\n-/// On Unix, `ssize_t` equals `isize`. On 64-bit Windows MSVC, `ssize_t`\n-/// is defined... | 2026-03-23T07:15:02 |
ollama/ollama | 515f497e6d5c0a102f913693fc8d4ed5eefd4ebb | b27268aaefa03f401f0f09c1e21b63872288ddc7 | fix: skip removing layers that no longer exist | [
{
"path": "server/manifest.go",
"patch": "@@ -3,6 +3,7 @@ package server\n import (\n \t\"crypto/sha256\"\n \t\"encoding/json\"\n+\t\"errors\"\n \t\"fmt\"\n \t\"io\"\n \t\"log/slog\"\n@@ -43,7 +44,9 @@ func (m *Manifest) Remove() error {\n \n func (m *Manifest) RemoveLayers() error {\n \tfor _, layer := ran... | 2024-06-10T18:15:03 |
ggml-org/llama.cpp | 23f82f242072ba56db8335484f8fe8bcb9208feb | 2656c0d2652964a66012426f5ec47b283a094cc5 | preset: allow named remote preset (#18728)
* preset: allow named remote preset
* nits: fix docs
* cont docs | [
{
"path": "common/arg.cpp",
"patch": "@@ -281,12 +281,20 @@ static std::string clean_file_name(const std::string & fname) {\n static bool common_params_handle_remote_preset(common_params & params, llama_example ex) {\n GGML_ASSERT(!params.model.hf_repo.empty());\n \n+ // the returned hf_repo is witho... | 2026-01-10T14:12:29 |
huggingface/transformers | 5d2f82b530d9e53fdb64b4e78638cd8f1673378f | 4e7cecb24d5ee9e41e488bebba69151faf6e3e4d | Fix GraniteMoeHybrid in transformers v5 (#42872)
* apply_rotary_pos_emb should be called
* fix position_embeddings usage in granitemoehybrid
* setting `self.rotary_emb` to None only in hybrid models. Safer, since all modules are highly modular.
* minor
* adding `position_embedding_type` to the config.
* review cl... | [
{
"path": "src/transformers/models/granitemoehybrid/configuration_granitemoehybrid.py",
"patch": "@@ -92,6 +92,8 @@ class GraniteMoeHybridConfig(PreTrainedConfig):\n allow the model to output the auxiliary loss.\n router_aux_loss_coef (`float`, *optional*, defaults to 0.001): router auxi... | 2025-12-16T11:18:00 |
vuejs/vue | 0d4b35f55975946cb0eb4f7f5f35efe3d078473e | 5d721a42b140865e50a78445fe21c5f270bde703 | fix(vdom): remove unnecessary sameVnode condition
fix #9168 | [
{
"path": "src/core/vdom/patch.js",
"patch": "@@ -32,20 +32,13 @@ export const emptyNode = new VNode('', {}, [])\n \n const hooks = ['create', 'activate', 'update', 'remove', 'destroy']\n \n-function childrenIgnored (vnode) {\n- return vnode && vnode.data && vnode.data.domProps && (\n- vnode.data.domPro... | 2018-12-09T21:21:22 |
denoland/deno | ceb6936a983904511867d69986203786b68b7e21 | f5014bb6c3ce4d398b920fc9f0b492dead2019e2 | fix(ext/node): return all own properties from VM context property enumerator (#32911)
## Summary
- VM context's property enumerator callbacks were using
`GetPropertyNamesArgs::default()` which only returns enumerable,
non-symbol properties
- This caused `Reflect.ownKeys()`, `Object.getOwnPropertyNames()`, and
`Object... | [
{
"path": "ext/node/ops/vm.rs",
"patch": "@@ -1027,9 +1027,13 @@ fn property_enumerator<'s>(\n };\n \n let context_scope = &mut v8::ContextScope::new(scope, context);\n- let Some(properties) = sandbox\n- .get_property_names(context_scope, v8::GetPropertyNamesArgs::default())\n- else {\n+ let args ... | 2026-03-23T07:10:48 |
ggml-org/llama.cpp | 9ac2693a302adab4b5b44fb1fef52c9e6d14a770 | a61c8bc3bfae4f86b8205535bcea73f476b28c2c | server: fix n_cmpl not skipping processing prompt (#18663)
* server: fix n_cmpl not skipping processing
* fix infinite loop on empty batch
* cont : init child samplers + modify child logic
* cont : cleanup
* cont : improve n_cmpl logic
- launch the parent task first so it finds the slot with best cache
- parent t... | [
{
"path": "tools/server/server-context.cpp",
"patch": "@@ -79,6 +79,8 @@ struct server_slot {\n \n common_speculative * spec = nullptr;\n \n+ // TODO: move members that belong to the task (such as `generated_text`, `has_new_line`) to task_results_state\n+ // see https://github.com/ggml-org/l... | 2026-01-09T23:00:41 |
huggingface/transformers | 4e7cecb24d5ee9e41e488bebba69151faf6e3e4d | 252afd8968a3e0d6b755d0ee1295593c6e0b7927 | [`Ernie 4.5 Moe`] Fix routing, weights, and update expectations (#42653)
* fix
* fix dtype casting
* add "small" model test
* update for a10
* fix
* crap forgot
* proper decorator usage
* add comment
* style
* fix expectation from dtype casting | [
{
"path": "src/transformers/conversion_mapping.py",
"patch": "@@ -166,6 +166,9 @@ def _build_checkpoint_conversion_mapping():\n mapping[\"deepseek_v3\"] = mapping[\"qwen2_moe\"].copy()\n mapping[\"dots1\"] = mapping[\"qwen2_moe\"].copy()\n mapping[\"ernie4_5_moe\"] = mapping[\"qwen2_moe\"].copy(... | 2025-12-16T11:07:52 |
vuejs/vue | 5d721a42b140865e50a78445fe21c5f270bde703 | 780dac561b9cd6c3cec28f154f76e7d28352ebf3 | revert: fix(sfc): avoid deindent when pad option is specified (#7647)
This reverts commit 9d2f9a034f9c40d5ba6d8b1e131b1bfb675dc1cf. | [
{
"path": "src/sfc/parser.js",
"patch": "@@ -83,15 +83,11 @@ export function parseComponent (\n function end (tag: string, start: number) {\n if (depth === 1 && currentBlock) {\n currentBlock.end = start\n- let text = content.slice(currentBlock.start, currentBlock.end)\n+ let text = de... | 2018-12-09T20:52:17 |
denoland/deno | f5014bb6c3ce4d398b920fc9f0b492dead2019e2 | fd73dffa16d136cc55948da7804e6ad35ab145ed | fix(ext/node): child_process improvements (internalMessage, argv0, DEP0190) (#32885)
## Summary
Three child_process Node.js compatibility fixes:
### 1. Emit `internalMessage` event for `NODE_` prefixed IPC messages
In Node.js, IPC messages with a `cmd` property starting with `NODE_` are
internal messages emitted as... | [
{
"path": "ext/node/polyfills/internal/child_process.ts",
"patch": "@@ -71,6 +71,7 @@ import { Pipe, socketType } from \"ext:deno_node/internal_binding/pipe_wrap.ts\";\n import { Server as NetServer, Socket } from \"node:net\";\n import { Socket as DgramSocket } from \"node:dgram\";\n import {\n+ kArgv0,\n... | 2026-03-23T07:08:51 |
huggingface/transformers | 252afd8968a3e0d6b755d0ee1295593c6e0b7927 | 89998bddcab4e3d5ee03ed9fa1bf1a9339584709 | Fix convert_tekken_tokenizer (#42592)
* Fix convert_tekken_tokenizer
* quality
---------
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com> | [
{
"path": "src/transformers/integrations/mistral.py",
"patch": "@@ -77,6 +77,7 @@ def convert_tekken_tokenizer(tokenizer_file: str):\n \"\"\"Convert a \"tekken\" tokenizer to a fast Tokenizer.\"\"\"\n # Tekken format -- need to use the Converter\n \n+ from mistral_common.tokens.tokenizers.base im... | 2025-12-16T11:03:29 |
vuejs/vue | 780dac561b9cd6c3cec28f154f76e7d28352ebf3 | a89384c8d93a3e5a382f60a84d60f0675408ed75 | fix(ssr): should not warn for custom directives that do not have ssr implementation
fix #9167 | [
{
"path": "src/server/render.js",
"patch": "@@ -326,7 +326,7 @@ function renderStartingTag (node: VNode, context) {\n for (let i = 0; i < dirs.length; i++) {\n const name = dirs[i].name\n if (name !== 'show') {\n- const dirRenderer = resolveAsset(context, 'directives', name, t... | 2018-12-09T20:47:43 |
ggml-org/llama.cpp | a61c8bc3bfae4f86b8205535bcea73f476b28c2c | 593da7fa49503b68f9f01700be9f508f1e528992 | mtmd: Add Gemma3n multimodal support with MobileNetV5 vision encoder (#18256)
* Add Gemma3nVisionModel - MobileNetV5 vision encoder convertor to convert_hf_to_gguf.py. Add gemma3n to vision projectors in gguf-py/gguf/constants.py.
* Add mobilenetv5 impl
* Fix comments, remove unused vars
* Fix permute and remove tr... | [
{
"path": "convert_hf_to_gguf.py",
"patch": "@@ -528,7 +528,11 @@ def generate_extra_tensors(self) -> Iterable[tuple[str, Tensor]]:\n return ()\n \n def prepare_tensors(self):\n- max_name_len = max(len(s) for _, s in self.tensor_map.mapping.values()) + len(\".weight,\")\n+ # Handle... | 2026-01-09T22:42:38 |
denoland/deno | fd73dffa16d136cc55948da7804e6ad35ab145ed | ac659a17c799c0e03530acd73524596f62f108fd | fix(ext/console): handle throwing Symbol.toStringTag getter gracefully (#32917)
## Summary
When `console.log` formats an object, it accesses `Symbol.toStringTag`
which may be a getter that throws. For example,
[quick-lru](https://www.npmjs.com/package/quick-lru)'s
`Symbol.toStringTag` getter calls `JSON.stringify(thi... | [
{
"path": "ext/web/01_console.js",
"patch": "@@ -714,7 +714,11 @@ function formatRaw(ctx, value, recurseTimes, typedArray, proxyDetails) {\n \n let tag;\n if (!proxyDetails) {\n- tag = value[SymbolToStringTag];\n+ try {\n+ tag = value[SymbolToStringTag];\n+ } catch {\n+ // Symbol.toSt... | 2026-03-23T07:07:10 |
huggingface/transformers | 8d526c238ac1d32b181b62b564b90149be38e6f9 | 7960b5ea40a310fdd55ec8eadbc38671965f6e42 | [modular] Fix a weird renaming edge-case (#42844)
* clean fix
* style | [
{
"path": "examples/modular-transformers/modeling_test_suffix.py",
"patch": "@@ -0,0 +1,250 @@\n+# 🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨🚨\n+# This file was automatically generated from examples/modular-transformers/modular_te... | 2025-12-16T10:40:20 |
vuejs/vue | dfaf9e24361e10ae68ce3951eaf48262cf90f0ec | 8a2dbf50105ea729125a42fecfe2c2f0371d7836 | fix(types): type support for advanced async components (#8438) | [
{
"path": "types/options.d.ts",
"patch": "@@ -15,12 +15,24 @@ export type Component<Data=DefaultData<never>, Methods=DefaultMethods<never>, Co\n interface EsModuleComponent {\n default: Component\n }\n+ \n+export type AsyncComponent<Data=DefaultData<never>, Methods=DefaultMethods<neve... | 2018-12-05T23:00:56 |
ggml-org/llama.cpp | ec8fd7876b6195bd8582eba4f1debb23d13fde81 | a180ba78c710f52af7c7edee562fbe98c36d6ddd | Webui/file upload (#18694)
* webui: fix restrictive file type validation
* webui: simplify file processing logic
* chore: update webui build output
* webui: remove file picker extension whitelist (1/2)
* webui: remove file picker extension whitelist (2/2)
* chore: update webui build output
* refactor: Cleanup
*... | [
{
"path": "tools/server/webui/src/lib/components/app/chat/ChatForm/ChatForm.svelte",
"patch": "@@ -10,21 +10,11 @@\n \timport { INPUT_CLASSES } from '$lib/constants/input-classes';\n \timport { SETTING_CONFIG_DEFAULT } from '$lib/constants/settings-config';\n \timport { config } from '$lib/stores/settings.s... | 2026-01-09T15:45:32 |
denoland/deno | ac659a17c799c0e03530acd73524596f62f108fd | eb108d2b4d722138700f33f8a6840a51281ed340 | fix(ext/node): do not throw error for `readline.createInterface` outside TTY (#32874)
Matches Node's behaviour by ignoring `terminal: true` when not inside a
TTY.
Fixes #32873.
---------
Co-authored-by: Bartek Iwańczuk <biwanczuk@gmail.com>
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com> | [
{
"path": "ext/node/polyfills/_process/streams.mjs",
"patch": "@@ -346,7 +346,9 @@ export const initStdin = (warmup = false) => {\n });\n stdin._isRawMode = false;\n stdin.setRawMode = (enable) => {\n- io.stdin?.setRaw?.(enable);\n+ if (io.stdin?.isTerminal()) {\n+ io.stdin.setRaw(enable);\... | 2026-03-22T08:30:19 |
ollama/ollama | 94d37fdcae30ddeb6c9f65c8707004f5ec9eaf33 | b84aea1685329f107c5547b1cca4efcf15c19f52 | fix: examples/langchain-python-rag-privategpt/requirements.txt (#3382) | [
{
"path": "examples/langchain-python-rag-privategpt/requirements.txt",
"patch": "@@ -11,4 +11,5 @@ tabulate==0.9.0\n pandoc==2.3\n pypandoc==1.11\n tqdm==4.66.1\n-sentence_transformers==2.2.2\n\\ No newline at end of file\n+sentence_transformers==2.2.2\n+numpy>=1.22.2 # not directly required, pinned by Snyk... | 2024-06-09T17:58:09 |
vuejs/vue | 8a2dbf50105ea729125a42fecfe2c2f0371d7836 | 0ed0aad77228b95e9a61a87386736938837527f8 | fix(transition-group): fix activeInstance regression
fix #9151 | [
{
"path": "src/core/instance/lifecycle.js",
"patch": "@@ -21,6 +21,14 @@ import {\n export let activeInstance: any = null\n export let isUpdatingChildComponent: boolean = false\n \n+export function setActiveInstance(vm: Component) {\n+ const prevActiveInstance = activeInstance\n+ activeInstance = vm\n+ r... | 2018-12-05T22:33:39 |
huggingface/transformers | 6c7c992fafcfabf5832c185faa9e26b34bb4a2fd | 0f97c688d53e38967d53189f79e7b9d2ebf02282 | Add missing ModelOutput subclass return type hints (#41219)
* Add missing ModelOutput subclass return type hints
* Fix incorrect type hint: FlavaOutput (nn.Module subclass) -> FlavaModelOutput | [
{
"path": "src/transformers/models/dac/modeling_dac.py",
"patch": "@@ -16,7 +16,7 @@\n \n import math\n from dataclasses import dataclass\n-from typing import Optional\n+from typing import Optional, Union\n \n import numpy as np\n import torch\n@@ -583,7 +583,7 @@ def encode(\n input_values: torch.T... | 2025-12-16T10:32:06 |
ggml-org/llama.cpp | 53eb9435da3affa12a38a8b0fb29081698a8d1cc | d3435efc8a2f926f1d7a5f2a11ac161b798c355a | server : fix timing of prompt/generation (#18713) | [
{
"path": "tools/server/server-context.cpp",
"patch": "@@ -2615,10 +2615,6 @@ struct server_context_impl {\n // on successful decode, restore the original batch size\n n_batch = llama_n_batch(ctx);\n \n- // technically, measuring the time here excludes the sampling time fo... | 2026-01-09T10:59:50 |
ollama/ollama | b84aea1685329f107c5547b1cca4efcf15c19f52 | 896495de7b5814cca32ba83d2e9f2bf176ba98c8 | Critical fix from llama.cpp JSON grammar to forbid un-escaped escape characters inside strings, which breaks parsing. (#3782) | [
{
"path": "llm/server.go",
"patch": "@@ -606,7 +606,7 @@ array ::=\n \n string ::=\n \"\\\"\" (\n- [^\"\\\\] |\n+ [^\"\\\\\\x7F\\x00-\\x1F] |\n \"\\\\\" ([\"\\\\/bfnrt] | \"u\" [0-9a-fA-F] [0-9a-fA-F] [0-9a-fA-F] [0-9a-fA-F]) # escapes\n )* \"\\\"\" ws\n ",
"additions": 1,
"deletions"... | 2024-06-09T17:57:09 |
denoland/deno | eb108d2b4d722138700f33f8a6840a51281ed340 | fe10f95260be6547159522e0d1ac2a70570dd449 | chore: update deno_doc crate (#32896)
Major restructuring of the DocNode format (v2), addressing long-standing
design issues with how documentation nodes are modeled and serialized.
this also bumps the JSON schema version to 2.
The output now is a `Record<string, Document>` instead of `DocNode[]`.
For a complete view... | [
{
"path": "Cargo.lock",
"patch": "@@ -2254,9 +2254,9 @@ dependencies = [\n \n [[package]]\n name = \"deno_doc\"\n-version = \"0.194.1\"\n+version = \"0.197.0\"\n source = \"registry+https://github.com/rust-lang/crates.io-index\"\n-checksum = \"36eede374a5ecf0a6d83ecfc47d8878e35d7e3210de6e9ca18554402a5327114... | 2026-03-21T22:02:26 |
vuejs/vue | 0ed0aad77228b95e9a61a87386736938837527f8 | f077ed17af14bb2675db64b2aa2d023769219624 | fix: fix keyName checking for space and delete in IE11 (#9150)
fix #9112 | [
{
"path": "src/compiler/codegen/events.js",
"patch": "@@ -22,13 +22,15 @@ const keyNames: { [key: string]: string | Array<string> } = {\n esc: ['Esc', 'Escape'],\n tab: 'Tab',\n enter: 'Enter',\n- space: ' ',\n+ // #9112: IE11 uses `Spacebar` for Space key name.\n+ space: [' ', 'Spacebar'],\n // ... | 2018-12-05T20:26:58 |
huggingface/transformers | 0f97c688d53e38967d53189f79e7b9d2ebf02282 | 7f52a2a4ea8ab49b7f069df7fac58a5b280d4919 | Fix BLT training_ci overfit test (#42685)
* Fix BLT training_ci overfit test by disabling cache and adjusting training thresholds
* Fix BLT training_ci overfit test by disabling cache and adjusting training thresholds
* Fix BLT training_ci overfit test by disabling cache and adjusting training thresholds
* Format B... | [
{
"path": "src/transformers/models/blt/modeling_blt.py",
"patch": "@@ -444,6 +444,155 @@ class BltPreTrainedModel(PreTrainedModel):\n \"attentions\": OutputRecorder(BltSelfAttention, index=1, layer_name=\"local_decoder\"),\n }\n \n+ @torch.no_grad()\n+ def _init_weights(self, module):\n+ ... | 2025-12-16T10:18:49 |
ggml-org/llama.cpp | 8ece3836b400dd8d89021ad2cc6e57843ced8378 | 046d5fd44e3505ab9c6d065ab65541fc2fdfd4f2 | common: support remote preset (#18520)
* arg: support remote preset
* proof reading
* allow one HF repo to point to multiple HF repos
* docs: mention about multiple GGUF use case
* correct clean_file_name
* download: also return HTTP status code
* fix case with cache file used
* fix --offline option | [
{
"path": "common/arg.cpp",
"patch": "@@ -6,6 +6,7 @@\n #include \"log.h\"\n #include \"sampling.h\"\n #include \"download.h\"\n+#include \"preset.h\"\n \n // fix problem with std::min and std::max\n #if defined(_WIN32)\n@@ -268,6 +269,46 @@ static void parse_tensor_buffer_overrides(const std::string & valu... | 2026-01-08T21:35:40 |
ollama/ollama | 5528dd9d1170e7a78a4fdb7684e8944e2052ca8f | 943172cbf4d6cc0b8682021bfc9c2d816152615d | Error handling load_single_document() in ingest.py (#4852)
load_single_document() handles
- corrupt files
- empty (zero byte) files
- unsupported file extensions | [
{
"path": "examples/langchain-python-rag-privategpt/ingest.py",
"patch": "@@ -77,13 +77,21 @@ def load(self) -> List[Document]:\n \n \n def load_single_document(file_path: str) -> List[Document]:\n- ext = \".\" + file_path.rsplit(\".\", 1)[-1]\n- if ext in LOADER_MAPPING:\n- loader_class, loade... | 2024-06-09T17:41:07 |
denoland/deno | 8f17347a5d05ee27be9adef2349c56a88bd0f199 | bac58b43ffaf3917d42805df68119486d2047461 | fix: skip flaky test-process-threadCpuUsage-worker-threads on Windows (#32886)
Skip `parallel/test-process-threadCpuUsage-worker-threads.js` on Windows
CI where it is flaky.
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com> | [
{
"path": "tests/node_compat/config.jsonc",
"patch": "@@ -1700,7 +1700,10 @@\n \"parallel/test-process-title-cli.js\": {},\n \"parallel/test-process-release.js\": {},\n \"parallel/test-process-threadCpuUsage-main-thread.js\": {},\n- \"parallel/test-process-threadCpuUsage-worker-threads.js\": ... | 2026-03-21T09:46:40 |
vuejs/vue | f077ed17af14bb2675db64b2aa2d023769219624 | 93850f49e31559c78dc91f5df9e756f424ba910a | fix(ssr): fix ssr template publicPath generation
fix #9145 | [
{
"path": "src/server/template-renderer/index.js",
"patch": "@@ -60,7 +60,10 @@ export default class TemplateRenderer {\n // extra functionality with client manifest\n if (options.clientManifest) {\n const clientManifest = this.clientManifest = options.clientManifest\n- this.publicPath = ... | 2018-12-05T15:54:42 |
huggingface/transformers | 23394cc49177f1acdf20b0d840917b797b731d3c | 06378d40e613db0dfbfb79a5569017bae83d3929 | Simplify using custom resolution for sam3 and sam3_video inference (#42787)
* simplify using custom resolution for sam3 and sam3_video inference
* revert auto format
* use setters and properties
* Fix docstring
* update dict to correctly save image_size to file for backward compatibility | [
{
"path": "docs/source/en/model_doc/sam3.md",
"patch": "@@ -354,6 +354,21 @@ When running the same text prompt on multiple images, pre-compute text embedding\n ... print(f\"Image {i+1}: {len(results['masks'])} '{text_prompt}' objects found\")\n ```\n \n+### Custom Resolution Inference\n+\n+<div class=\"... | 2025-12-15T19:44:34 |
ggml-org/llama.cpp | 480160d47297df43b43746294963476fc0a6e10f | 15bff84bf56651d6f991f166a2bf0f362996f7f9 | ggml-webgpu: Fix GGML_MEM_ALIGN to 8 for emscripten. (#18628)
* Fix GGML_MEM_ALIGN to 8 for emscripten.
* Add a comment explaining the need for GGML_MEM_ALIGN == 8 in 64-bit wasm with emscripten | [
{
"path": "ggml/include/ggml.h",
"patch": "@@ -234,6 +234,11 @@\n \n #if UINTPTR_MAX == 0xFFFFFFFF\n #define GGML_MEM_ALIGN 4\n+#elif defined(__EMSCRIPTEN__)\n+// emscripten uses max_align_t == 8, so we need GGML_MEM_ALIGN == 8 for 64-bit wasm.\n+// (for 32-bit wasm, the first conditional is true and GG... | 2026-01-08T16:36:42 |
ollama/ollama | 34f142797aff17af2a1c22d68529269d231cc8d4 | 46a7f1e74ace3425254ac9cfba2020026c3427e3 | llm: always add bos token to prompt (#4941)
* fix embedding by adding fixes from llama.cpp upstream
* remove assert
---------
Co-authored-by: Jesper Ek <deadbeef84@gmail.com> | [
{
"path": "llm/ext_server/server.cpp",
"patch": "@@ -835,7 +835,7 @@ struct llama_server_context\n system_tokens.clear();\n \n if (!system_prompt.empty()) {\n- system_tokens = ::llama_tokenize(ctx, system_prompt, add_bos_token);\n+ system_tokens = ::llama_tokenize(ctx, ... | 2024-06-09T01:47:10 |
vuejs/vue | 93850f49e31559c78dc91f5df9e756f424ba910a | 1b4a8a0c1edaf9c7eb129ba61bca94ba607bbf56 | chore: fix sponsor link | [
{
"path": "BACKERS.md",
"patch": "@@ -195,8 +195,8 @@ Funds donated via Patreon go directly to support Evan You's full-time work on Vu\n </a>\n </td>\n <td align=\"center\" valign=\"middle\">\n- <a href=\"https://www.inkloop.in/\" target=\"_blank\">\n- <img width=\"148px\... | 2018-12-04T16:00:09 |
denoland/deno | 8b14d5afdce7ab928a8b2ae2ac9147d9d913ec25 | 38186cc4f5024d57a7eb9c4ba93b9062114dd8f0 | fix(audit): respect package.json overrides in vulnerability checks (#32878)
## Summary
Closes #32871
When `package.json` overrides force a transitive dependency to a patched
version, `deno audit` was still reporting it as vulnerable. For example:
```json
{
"dependencies": { "@vitest/ui": "^4.1.0" },
"overrides"... | [
{
"path": "cli/tools/pm/audit.rs",
"patch": "@@ -318,6 +318,49 @@ mod npm {\n \n let mut advisories = response.advisories.values().collect::<Vec<_>>();\n \n+ // Filter out advisories where no installed version falls within\n+ // the vulnerable range. This handles package.json overrides that\n+ ... | 2026-03-21T09:01:26 |
huggingface/transformers | 06378d40e613db0dfbfb79a5569017bae83d3929 | fc50bdc68529f05b5f571ef8aaa228698ea9b3f4 | fix: Initialize ApertusMLP's xielu activation using `torch_dtype` (#42864)
* Fix Apertus model crash on float16 hardware
Initialize XIELU activation with correct dtype from config (using config.dtype instead of default bfloat16) to prevent promotion to float32 and subsequent crashes on Turing/float16 GPUs.
* refacto... | [
{
"path": "src/transformers/models/apertus/modeling_apertus.py",
"patch": "@@ -25,7 +25,7 @@\n import torch\n from torch import nn\n \n-from ...activations import ACT2FN\n+from ...activations import ACT2CLS, ACT2FN\n from ...cache_utils import Cache, DynamicCache\n from ...generation import GenerationMixin\... | 2025-12-15T16:09:49 |
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.