Repository Issue Activity (beta)

#1411 Add fidelity-to-teacher KLD and long-context retrieval eval types

#1433 Improve QAIRT model performance on Snapdragon X Plus 10-Core machines

#1397 Build Qwen 3.8 for qairt and get it on the AI Hub website ASAP

#1432 Android JNI: applyChatTemplate failure is returned as an empty formattedText instead of an error

#1436 fix(sdk): initialize VLM encoder from resolved devices

#1434 Generation error error="SDKError(Multimodal generation failed)"` with Gemma4 models

#1379 can't run `unsloth/Qwen3.8-27B-GGUF` MTP Q4

#1431 Qualcomm Adreno OpenCL driver (CB.dll) crashes llama_cpp/GGUF serving with 0xC0000409

#1429 ggml-hexagon on Hexagon v73 (QCM6690): NPU output is byte-identical regardless of the prompt — silent corruption, no crash, `graph splits = 3`

#1244 IQ9-EVK NPU Limitations?

#1286 Image Generation Support with NPU

#1403 geniex serve: max_tokens ignored on the qairt plugin — always generates to a fixed 2048-token cap

#1085 ModelScope integration status — confirm whether to re-introduce it

#1413 Add transaction-safe lineage validation for retained chat caches

#1409 # Feature request: MoE expert streaming / caching to run large MoE models

#1070 Add an Eval harness

#1201 VLM gibberish

#1398 Genie-X dll is being blocked by Windows 11 Smart App Control

#1314 I made a UI Version how can I publish it?

#1395 GenieX support to run T5 model on NPU

#1390 ggml-hexagon NPU backend: remote session open fails on Snapdragon X Elite WoA (HND driver v1.0.0.14) — GGML_ASSERT(device) in v0.5.0 NPU inference

#1380 Support Speech Recongition Models

#1153 GenieX can support Snapdragon X Plus 10-Core

#1170 ggml-hexagon: QCS6490 (Hexagon V68) crashes during prompt prefill with dspqueue_read failed: 0x0000002e

#1341 /v1/completions fails on QAIRT models because the qairt plugin rejects stop sequences