Repository navigation

Issues

New issue

Transformers' version 5 is out!
#43489 · LysandreJik opened on Jan 26, 2026
8
Tell Us: What Would Make Trainer Better?
#43595 · SunMarc opened on Jan 29, 2026
16

Generation crashes with encoder_repetition_penalty when encoder_outputs are provided without input_ids
huggingface/transformers#49695 · opened on Oct 17, 2026
2
[serge] integration failure triage - 2026-10-16
huggingface/transformers#49683 · opened on Oct 16, 2026
TextStreamer re-decodes the whole line on every token (quadratic on long lines)
huggingface/transformers#49675 · opened on Oct 16, 2026
Chat parsing: json parser's string_delims doesn't handle escaped delimiters
huggingface/transformers#49674 · opened on Oct 16, 2026
1
Quick survey: your experience as a PyTorch Ecosystem maintainer
huggingface/transformers#49672 · opened on Oct 16, 2026
AutoTokenizer.from_pretrained deep-copies the freshly loaded backend for TokenizersBackend tokenizers
huggingface/transformers#49666 · opened on Oct 16, 2026
GGUF tokenizer ignores tokenizer.ggml.add_bos_token, so the default encoding drops BOS
huggingface/transformers#49647 · opened on Oct 15, 2026
BatchEncoding.convert_to_tensors copies the whole batch to check if it's empty (~2x slower collation)
huggingface/transformers#49645 · opened on Oct 15, 2026
2
Gemma4-12B Training Failed
huggingface/transformers#49630 · opened on Oct 15, 2026
1
[i18n-ur] Translating docs to Urdu (اردو)
huggingface/transformers#49626 · opened on Oct 15, 2026
1
Janus / DeepSeek-VL image processors ignore background_color; deepseek-vl-7b-chat's low-res branch pads with 127 instead of its saved (122, 116, 104)
huggingface/transformers#49625 · opened on Oct 15, 2026
3
_patch_mistral_regex treats a missing transformers_version as "Mistral": false positive on local DeBERTa-v3, and following its advice breaks the tokenizer
huggingface/transformers#49623 · opened on Oct 15, 2026
1
DebertaV2Tokenizer drops SentencePiece NFKC normalization in v5: ligatures, fullwidth and circled characters encode to [UNK]
huggingface/transformers#49622 · opened on Oct 15, 2026
4
SAM 3: text-only prompts drop the geometry CLS token that Meta's implementation always adds
huggingface/transformers#49592 · opened on Oct 14, 2026
3
import transformers spends 1-3 s in an eager importlib.metadata.packages_distributions() when torch >= 2.14 is installed (Python 3.12)
huggingface/transformers#49591 · opened on Oct 14, 2026
3
Tokenizer loses its normalizer from tokenizer.json in 5.18
huggingface/transformers#49578 · opened on Oct 13, 2026
SDPA: enable_gqa=True silently falls back to the math kernel at batch 1 on torch < 2.14 without Flash Attention
huggingface/transformers#49564 · opened on Oct 13, 2026
2
/v1/systemone: 500s for chat_template_kwargs overrides and for models without a chat template
huggingface/transformers#49553 · opened on Oct 13, 2026
1
CPU SDPA could use native GQA with a mask on torch >= 2.9 (1.2x decode, no repeat_kv)
huggingface/transformers#49546 · opened on Oct 12, 2026
3
Gemma 3/4 multimodal models fail with tp_plan="auto" because of mixed Tensor/DTensor usage
huggingface/transformers#49540 · opened on Oct 12, 2026
4
Mamba2: non-fused gated RMSNorm ignores n_groups and disagrees with fused training
huggingface/transformers#49536 · opened on Oct 12, 2026
MM Grounding DINO and LLMDet: decoder box and class heads are tied to layer 0 although the config says they are not shared
huggingface/transformers#49529 · opened on Oct 12, 2026
Assigning rope_scaling drops existing RoPE base parameters
huggingface/transformers#49519 · opened on Oct 12, 2026
SynthID-Text: CPU and MPS produce different sampling tables with the same seed
huggingface/transformers#49498 · opened on Oct 11, 2026
4
ZeRO-3: from_pretrained reads and converts the full checkpoint on every rank
huggingface/transformers#49497 · opened on Oct 11, 2026
1

Navigation Menu

Repository navigation

Issues

Issues

Generation crashes with encoder_repetition_penalty when encoder_outputs are provided without input_ids

[serge] integration failure triage - 2026-10-16

TextStreamer re-decodes the whole line on every token (quadratic on long lines)

Chat parsing: `json` parser's `string_delims` doesn't handle escaped delimiters

Quick survey: your experience as a PyTorch Ecosystem maintainer

AutoTokenizer.from_pretrained deep-copies the freshly loaded backend for TokenizersBackend tokenizers

GGUF tokenizer ignores tokenizer.ggml.add_bos_token, so the default encoding drops BOS

BatchEncoding.convert_to_tensors copies the whole batch to check if it's empty (~2x slower collation)

Gemma4-12B Training Failed

[i18n-ur] Translating docs to Urdu (اردو)

Janus / DeepSeek-VL image processors ignore `background_color`; deepseek-vl-7b-chat's low-res branch pads with 127 instead of its saved (122, 116, 104)

_patch_mistral_regex treats a missing transformers_version as "Mistral": false positive on local DeBERTa-v3, and following its advice breaks the tokenizer

DebertaV2Tokenizer drops SentencePiece NFKC normalization in v5: ligatures, fullwidth and circled characters encode to [UNK]

SAM 3: text-only prompts drop the geometry CLS token that Meta's implementation always adds

import transformers spends 1-3 s in an eager importlib.metadata.packages_distributions() when torch >= 2.14 is installed (Python 3.12)

Tokenizer loses its normalizer from tokenizer.json in 5.18

SDPA: enable_gqa=True silently falls back to the math kernel at batch 1 on torch < 2.14 without Flash Attention

`/v1/systemone`: 500s for `chat_template_kwargs` overrides and for models without a chat template

CPU SDPA could use native GQA with a mask on torch >= 2.9 (1.2x decode, no repeat_kv)

Gemma 3/4 multimodal models fail with tp_plan="auto" because of mixed Tensor/DTensor usage

Mamba2: non-fused gated RMSNorm ignores n_groups and disagrees with fused training

MM Grounding DINO and LLMDet: decoder box and class heads are tied to layer 0 although the config says they are not shared

Assigning rope_scaling drops existing RoPE base parameters

SynthID-Text: CPU and MPS produce different sampling tables with the same seed

ZeRO-3: from_pretrained reads and converts the full checkpoint on every rank