Search results

[quantization] Implement tie_word_embeddings

Status: Open.

#624 In Samsung/TICO;

· stamalakhov opened on Apr 13, 2026

[quantization] Introduce Cross-Layer Equalization (CLE) algorithm

Status: Open.

#623 In Samsung/TICO;

· mhs4670go opened on Apr 13, 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

[LLama] Failed to evaluate truthfulqa

Status: Open.

#617 In Samsung/TICO;

· stamalakhov opened on Apr 10, 2026

Qwen3VL: Fixed Input Data Format At Inference Time

Status: Open.

#614 In Samsung/TICO;

· dvsav opened on Apr 7, 2026

[quanization] Enable prefill-decode modeling

Status: Open.

#586 In Samsung/TICO;

· stamalakhov opened on Mar 31, 2026

Introduce prefill_reference attention variant for faster PTQ evaluation

Status: Open.

#551 In Samsung/TICO;

· mhs4670go opened on Mar 13, 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

[quantization] Quantization-Aware Shape Operation Folding

Status: Open.

#499 In Samsung/TICO;

· dayo09 opened on Feb 19, 2026