-
Notifications
You must be signed in to change notification settings - Fork 27
All issues
Issue creation is restricted in this repository
Issues
is:issue state:open
is:issue state:open
Search results
feat(server): derive the thinking_mode chat-template kwarg when the template supports it
priority:lowLow priorityLow prioritystatus:readyReady to be worked onReady to be worked ontype:enhancementNew features, capabilities, or significant additionsNew features, capabilities, or significant additionsStatus: Open.#775 In lablup/mlxcel;feat(server): prompt-cache reject-reason metrics and opt-in trace logging
priority:lowLow priorityLow prioritystatus:readyReady to be worked onReady to be worked ontype:enhancementNew features, capabilities, or significant additionsNew features, capabilities, or significant additionsStatus: Open.#774 In lablup/mlxcel;fix(server): reject requests with no effective input before model dispatch
priority:mediumMedium priorityMedium prioritystatus:readyReady to be worked onReady to be worked ontype:bugBug fixes, error corrections, or issue resolutionsBug fixes, error corrections, or issue resolutionsStatus: Open.#773 In lablup/mlxcel;feat(sampling): implement XTC sampling (request params are accepted but ignored)
priority:lowLow priorityLow prioritystatus:readyReady to be worked onReady to be worked ontype:enhancementNew features, capabilities, or significant additionsNew features, capabilities, or significant additionsStatus: Open.#772 In lablup/mlxcel;feat(vlm): add Unlimited-OCR support (document OCR with ring-sliding decode cache)
priority:mediumMedium priorityMedium prioritystatus:reviewUnder reviewUnder reviewtype:enhancementNew features, capabilities, or significant additionsNew features, capabilities, or significant additionsStatus: Open.#771 In lablup/mlxcel;perf(server): bound the tick-slice speculative-slot hold so long streams cannot monopolize speculative acceleration
area:inferenceGeneration, sampling, decoding (incl. speculative, DRY)Generation, sampling, decoding (incl. speculative, DRY)priority:lowLow priorityLow prioritystatus:readyReady to be worked onReady to be worked ontype:performancePerformance improvementsPerformance improvementsStatus: Open.#746 In lablup/mlxcel;feat: extend the OpenXLA packed-quant path to bf16 scales/biases
area:inferenceGeneration, sampling, decoding (incl. speculative, DRY)Generation, sampling, decoding (incl. speculative, DRY)priority:lowLow priorityLow prioritystatus:readyReady to be worked onReady to be worked ontype:enhancementNew features, capabilities, or significant additionsNew features, capabilities, or significant additionsStatus: Open.#603 In lablup/mlxcel;track(upstream-iree): fused dequant->matmul for the packed int8 bandwidth win
area:inferenceGeneration, sampling, decoding (incl. speculative, DRY)Generation, sampling, decoding (incl. speculative, DRY)priority:lowLow priorityLow prioritystatus:blockedBlocked by dependencies or other issuesBlocked by dependencies or other issuestype:performancePerformance improvementsPerformance improvementsStatus: Open.#574 In lablup/mlxcel;epic: OpenXLA multimodal / VLM architecture support (Qwen2-VL, Gemma3n, Phi4MM, Molmo, and more)
area:architectureArchitecture and code structure changesArchitecture and code structure changesarea:inferenceGeneration, sampling, decoding (incl. speculative, DRY)Generation, sampling, decoding (incl. speculative, DRY)priority:lowLow priorityLow prioritystatus:readyReady to be worked onReady to be worked ontype:enhancementNew features, capabilities, or significant additionsNew features, capabilities, or significant additionsStatus: Open.#566 In lablup/mlxcel;epic: OpenXLA SSM / hybrid / recurrent architecture support (Mamba2, Nemotron-H, Jamba, and more)
area:architectureArchitecture and code structure changesArchitecture and code structure changesarea:inferenceGeneration, sampling, decoding (incl. speculative, DRY)Generation, sampling, decoding (incl. speculative, DRY)priority:lowLow priorityLow prioritystatus:readyReady to be worked onReady to be worked ontype:enhancementNew features, capabilities, or significant additionsNew features, capabilities, or significant additionsStatus: Open.#564 In lablup/mlxcel;feat(deepseek-v4): clamp shared-expert SwiGLU to swiglu_limit
area:modelsModel architectures, weights, loading, metadataModel architectures, weights, loading, metadatapriority:lowLow priorityLow prioritystatus:blockedBlocked by dependencies or other issuesBlocked by dependencies or other issuestype:enhancementNew features, capabilities, or significant additionsNew features, capabilities, or significant additionsStatus: Open.#550 In lablup/mlxcel;feat(deepseek-v4): implement HiSA hierarchical sparse attention
area:modelsModel architectures, weights, loading, metadataModel architectures, weights, loading, metadatapriority:lowLow priorityLow prioritystatus:blockedBlocked by dependencies or other issuesBlocked by dependencies or other issuestype:enhancementNew features, capabilities, or significant additionsNew features, capabilities, or significant additionsStatus: Open.#549 In lablup/mlxcel;