Search results

[quantization] Support Gemma4

Status: Open.

#768 In Samsung/TICO;

· mhs4670go opened on Jun 10, 2026

[examples] Refactor WrapQ examples into config-driven reusable recipe structure

Status: Open.

#716 In Samsung/TICO;

· mhs4670go opened on May 19, 2026

[quantization] Instability on llama quantization

Status: Open.

#656 In Samsung/TICO;

· stamalakhov opened on Apr 27, 2026

[quantization] Calibration Dataset Selection & Sampling Strategy

Status: Open.

#645 In Samsung/TICO;

· mhs4670go opened on Apr 22, 2026

[quantization] Make Attention Mask Fill Value Configurable & Run Ablation Study

Status: Open.

#643 In Samsung/TICO;

· mhs4670go opened on Apr 20, 2026

[quantization] Ablation study: float vs quantized prefill for decode calibration

Status: Open.

#622 In Samsung/TICO;

· mhs4670go opened on Apr 12, 2026

[quantization] Refactor: Remove float32 assumption and follow model dtype

Status: Open.

#621 In Samsung/TICO;

· mhs4670go opened on Apr 12, 2026

Qwen3VL: Fixed Input Data Format At Inference Time

Status: Open.

#614 In Samsung/TICO;

· dvsav opened on Apr 7, 2026

[wrapq] High PEIR (46%) of LlamaDecoderLayer of TinyLlama

Status: Open.

#513 In Samsung/TICO;

· dayo09 opened on Feb 24, 2026

[test] Install transformers for wrapq feature coverage

Status: Open.

#510 In Samsung/TICO;

· mhs4670go opened on Feb 24, 2026

[CI] Test wrapq examples

Status: Open.

#505 In Samsung/TICO;

· dayo09 opened on Feb 20, 2026

[wrapq] Operation Decomposition After Quantization

Status: Open.

#489 In Samsung/TICO;

· dayo09 opened on Feb 13, 2026