-
Notifications
You must be signed in to change notification settings - Fork 96
All issues
Issue creation is restricted in this repository
- #1213 · msluszniak opened
on Jun 1, 2026
Issues
is:issue state:open
is:issue state:open
Search results
Smarter repetition handling for small models: DRY, XTC and typical sampling
user expThis issue tackles problems with user experience e.g. overcomplicated APIThis issue tackles problems with user experience e.g. overcomplicated APIStatus: Open.Vulkan LLMs: what would make decode competitive
ideaNew idea to enhance the library, suggestion, etc.New idea to enhance the library, suggestion, etc.performanceRelated to all issues and tasks focused on improving performanceRelated to all issues and tasks focused on improving performanceplatform: androidIssues and tasks related to AndroidIssues and tasks related to AndroidStatus: Open.#1478 In software-mansion/react-native-executorch;Adopt the upstream split-backend linking once executorch#21849 lands
blockedIssue blocked by some problems (but not other issue, use relationship -> blocker instead)Issue blocked by some problems (but not other issue, use relationship -> blocker instead)executorchIssues and tasks related specifically to ExecuTorchIssues and tasks related specifically to ExecuTorchStatus: Open.#1474 In software-mansion/react-native-executorch;perf: sweep 0.9 vs 0.10 inference times, 24 of 42 model pairs are slower
modelIssues related to exporting, improving, fixing ML modelsIssues related to exporting, improving, fixing ML modelsperformanceRelated to all issues and tasks focused on improving performanceRelated to all issues and tasks focused on improving performanceStatus: Open.#1462 In software-mansion/react-native-executorch;Refactor LLM runner to bypass the upstream ET runner
ideaNew idea to enhance the library, suggestion, etc.New idea to enhance the library, suggestion, etc.Status: Open.Deprecate XNNPACK fp32 variants that have a faster, smaller quantized twin
improvementPRs or issues focused on improvements in the current codebasePRs or issues focused on improvements in the current codebasemodelIssues related to exporting, improving, fixing ML modelsIssues related to exporting, improving, fixing ML modelsperformanceRelated to all issues and tasks focused on improving performanceRelated to all issues and tasks focused on improving performanceStatus: Open.- Status: Open.
Support a larger context window
modelIssues related to exporting, improving, fixing ML modelsIssues related to exporting, improving, fixing ML modelsuser expThis issue tackles problems with user experience e.g. overcomplicated APIThis issue tackles problems with user experience e.g. overcomplicated APIStatus: Open.[RNE Rewrite] Drop the Android DownloadManager fallback once blob-util releases the #475 fix
3rd party packageIssue related to 3rd party packages, but not ExecuTorch, e.g. ExpoIssue related to 3rd party packages, but not ExecuTorch, e.g. Expoplatform: androidIssues and tasks related to AndroidIssues and tasks related to AndroidStatus: Open.#1401 In software-mansion/react-native-executorch;[RNE Rewrite] Rewrite React Native RAG package to new flow
chorePRs that are choresPRs that are choresStatus: Open.Try to export PrismML Bonsai models
ideaNew idea to enhance the library, suggestion, etc.New idea to enhance the library, suggestion, etc.modelIssues related to exporting, improving, fixing ML modelsIssues related to exporting, improving, fixing ML modelsStatus: Open.#1382 In software-mansion/react-native-executorch;MLX backend: large, deterministic logit divergence between Apple Silicon Mac and iPhone on identical inputs (~0.5x logit compression, ~18% answer flips)
pending responseThis issue awaits some actionThis issue awaits some actionplatform: iosIssues and tasks related to iOSIssues and tasks related to iOSuser expThis issue tackles problems with user experience e.g. overcomplicated APIThis issue tackles problems with user experience e.g. overcomplicated APIStatus: Open.