Pull Requests huggingface/sentence-transformers

fix typos: inaccuracte -> inaccurate

#3542 by whybe-choi was merged 2025-10-07 18:03

[`docs`] Format all markdown using mdformat

#3539 by tomaarsen was merged 2025-10-14 11:32

[`fix`] Allow MatryoshkaLoss with MSELoss

#3538 by tomaarsen was merged 2025-10-14 11:30

Add support for Intel 'xpu' device in environment.py

#3537 by domschl was merged 2025-10-21 10:57

fix: allow MSE to be used with matryoshka

#3535 by stephantul was closed 2025-10-06 15:51

Performance table format and Few markdownlint error correction

#3534 by kaushikacharya was merged 2025-10-03 12:31

Future-proof tests for upcoming transformers version

#3533 by tomaarsen was merged 2025-09-30 07:33

feat: matryoshka dims bounds checks and warning

#3530 by stephantul was merged 2025-09-30 07:33

[`tests`] Patch Regex expected output for Python 3.9

#3529 by tomaarsen was merged 2025-09-26 15:38

[`fix`] Allow loading Dense modules not saved in fp32

#3528 by tomaarsen was merged 2025-09-26 15:08

[`fix`] Patch Router training with Cached losses

#3527 by tomaarsen was merged 2025-09-26 14:49

bug: fix state bug in trainer

#3525 by stephantul was merged 2025-09-26 14:34

fix: static model saving fails because weights are not contiguous

#3524 by stephantul was merged 2025-09-24 07:28

Fix:Import SentenceTransformer class explicitly in losses module

#3521 by altescy was merged 2025-09-22 07:04

fix: add makedirs to informationretrievalevaluator

#3516 by stephantul was merged 2025-09-22 07:06

always pass `input_ids`, `attention_mask`, `token_type_ids`, `inputs_embeds` to forward

#3509 by Samoed was merged 2025-09-02 10:32

Update rasyosef/splade-mini MSMARCO and BEIR-13 benchmark scores in pretrained_models.md

#3508 by rasyosef was merged 2025-09-01 11:08

Mine hard negatives: optionally output similarity scores

#3506 by tsbalzhanov was merged 2025-12-11 11:12

[`fix`] Fix the number of missing negatives in mine_hard_negatives

#3504 by tomaarsen was merged 2025-09-22 07:17

[`tests`] Add hard negatives test showing multiple positives are correctly handled

#3503 by tomaarsen was merged 2025-08-27 14:56

[`feat`] Use encode_document and encode_query in mine_hard_negatives

#3502 by tomaarsen was merged 2025-08-27 14:57

[`model_card`] Don't override manually provided languages in model card

#3501 by tomaarsen was merged 2025-08-27 10:55

[`feat`] add get_model_kwargs method; throw error if unused kwarg is passed

#3500 by tomaarsen was merged 2025-09-15 11:55

[`fix`] add batch size parameter to model prediction in CrossEncoderRerankingEvaluator

#3497 by emapco was merged 2025-08-26 10:46

Docs Patch for AnglE and CoSENT Losses

#3496 by johneckberg was merged 2025-08-26 10:38

Add Support for Knowledgeable Passage Retriever (KPR)

#3495 by ikuyamada was merged 2025-08-29 11:31

[`fix`] Ensure multi-process embeddings are moved to CPU for concatenation

#3488 by tomaarsen was merged 2025-08-27 10:53

[`docs`] Update main sbert.net page with v5.1 mention

#3482 by tomaarsen was merged 2025-08-06 13:00

[`feat`] Allow n-tuples for CE MarginMSE training

#3481 by tomaarsen was merged 2025-08-06 12:53

[`fix`] FIPS compatibility - use SHA256 with usedforsecurity=False in hard negatives caching

#3479 by tomaarsen was merged 2025-08-06 08:25