Problemsby IbisLlama users keep filing bugs
LLM apps2 posts from 2 people2d active+22 posts in the last 7 days, 0 the 7 days before (new)Unverified
Posts per day
The posts behind it
2, newest firstInteger `seed` above 2^53 loses precision (options decoded as float64), so distinct seeds give identical output### What is the issue? Integer options in `options` are decoded from JSON as `float64` before they are converted to `int`. A `seed` larger than 2^53 is rounded to the nearest representable double. So **different seeds can become the same effective seed**, and the request still...ollama/ollama · bigblue-r42 repliesOpen ↗AutoTokenizer.from_pretrained deep-copies the freshly loaded backend for TokenizersBackend tokenizersFor tokenizers that load through `TokenizersBackend` (Hub repos with `tokenizer_class: PreTrainedTokenizerFast`, e.g. Llama 3, GLM-4.x, Granite, OLMo-2), `convert_to_native_format` builds the Rust tokenizer with `TokenizerFast.from_file` and passes it as `tokenizer_object`....huggingface/transformers · deadstrobe5Open ↗
Companies and products named
About this problem
Evidence
2 posts from 2 people in 2 places, about 2 a week over 2 days. Mostly on huggingface/transformers, ollama/ollama. Tools named alongside: Hugging Face, Ollama, GitHub.
Frustration Frustration 0 of 3· Seen on GitHub issues
How it was grouped
Posts about Llama that state a pain but match no known issue yet. First post Oct 16, 2026, latest Oct 17, 2026. Unverified: fewer than 3 posts, or one person or place only. Method
Other Llama problems
History
Llama users keep filing bugs (2 posts)