Problemsby GlyphTrouble with Hugging Face logits
4 posts from 4 people3d activeFading, +0 this week
Posts per day
The posts behind it
[Doge] `DogeCDMoE` router `.view(2, bsz * seq_len, -1)` mixes tokens: with `is_moe=True`, logits depend on future tokens and on other sequences in the batch
### System Info - `transformers` version: 5.22.1 - Platform: Linux-6.6.87.2-microsoft-standard-WSL2-x86_64-with-glibc2.41 - Python version: 3.13.13 - Huggingface_hub version: 2.4.0 - Safetensors version: 0.8.0 - Accelerate version: 1.15.0 - Accelerate config: not found -...
ModernBERT mean pooling overflows in float16 on long inputs and returns NaN logits
### System Info - `transformers` version: 5.22.1 - Platform: macOS-27.0.1-arm64-arm-64bit-Mach-O (Apple M5 Pro, run on CPU) - Python version: 3.13.15 - Huggingface_hub version: 2.4.0 - Safetensors version: 0.8.0 - Accelerate version: not installed - Accelerate config: not found...
Qwen3_5MoeModel no longer returns router_logits, breaking aux loss with output_router_logits=True
### System Info transformers 5.18.0.dev0 (main, after #45783) ### Who can help? @zucchini-nlp (author of #45783), @vasqu (who reviewed it) ### Reproduction ### Expected behavior `router_logits` should be returned and `aux_loss` should be a tensor, as before #45783. #45783...
Vectorized (batched) logits processors for assisted decoding
### Feature request Vectorize the logits processor application over draft slots in assisted decoding, so speculative candidate generation can apply processors to all drafted positions in one call instead of looping one draft step at a time. ### Context This came up during review...
Companies and products named
About this problem
Evidence
4 posts from 4 people in 1 place, about 1 a week over 66 days. Mostly on huggingface/transformers. Tools named alongside: PyTorch, GitHub, NVIDIA.
Frustration Frustration 0 of 3· Seen on GitHub issues
How it was grouped
Posts about Hugging Face with no known issue, sharing the word "logits". First post Aug 28, 2026, latest Nov 1, 2026. Corroborated: 3 or more posts from 2 or more people or places. Method
Other Hugging Face problems
Hugging Face upgrades break working setups
14 posts from 13 people on GitHub issues, Reddit, Community forums.
Running Hugging Face locally hits memory and speed limits
17 posts from 15 people on GitHub issues, Reddit.
Fine-tuning Hugging Face is costly and unpredictable
9 posts from 9 people on GitHub issues, Reddit.
Trouble with Hugging Face fail
Top post: "All paid GPU pools fail to start: MOUNT_ATTR_IDMAP error at container init (Exit code 128)", one of 8 posts from 5 people on GitHub issues, Community forums.
Trouble with Hugging Face generate
Top post: "prefill_chunk_size drops image inputs, VLMs generate without the image", one of 6 posts from 5 people on GitHub issues, Community forums.
Trouble with Hugging Face token
Top post: "<|image|> token has special=False in Rust tokenizer backend, should it be True?", one of 4 posts from 4 people on GitHub issues.
Trouble with Hugging Face docs
Top post: "Translating docs to <languageName>", one of 4 posts from 3 people on GitHub issues.
Trouble with Hugging Face torch
Top post: "import transformers spends 1-3 s in an eager importlib.metadata.packages_distributions() when torch >= 2.14", one of 4 posts from 4 people on GitHub issues.
Trouble with Hugging Face load
Top post: "gemma 4 can not load audio file and produce Audio features and audio tokens do not match, tokens: 31,", one of 3 posts from 3 people on GitHub issues.
Trouble with Hugging Face chat
Top post: "Chat parsing: `json` parser's `string_delims` doesn't handle escaped delimiters", one of 3 posts from 3 people on GitHub issues, Community forums.
Hugging Face runs out of context on real work
3 posts from 3 people on GitHub issues.
Trouble with Hugging Face size
Top post: "Inkling-Small: width of the routed experts resolves to 3072 instead of 2048 (moe_intermediate_size is never", one of 3 posts from 3 people on GitHub issues.
History