Rudra Murthy

> Before diving a bit deeper, I don't really understand why are you using `convert_id_to_tokens` instead of juste using the `tokenizer.batch_decode` method? Did you try with it? Hi @ArthurZucker ,...

Hi, It is not restricted to change in class names from `PositiveDefinite` with `SymmetricPositiveDefinite` . The signature of the class `Problem` is also changed. Old constructor used in GeoMM `...

Hi, any updates on this functionality?

Hi, I am getting a similar error as well. I am using version `0.4.3`. This is the command I am running ``` lm_eval --model vllm --model_args pretrained=sarvamai/meta-llama/Meta-Llama-3.1-8B-Instruct --tasks bbh_zeroshot_word_sorting --num_fewshot...

On looking into the code, I see the issue is with this function [here](https://github.com/EleutherAI/lm-evaluation-harness/blob/v0.4.3/lm_eval/api/registry.py#L188C1-L192C74). This expects a string as input and searches in the corresponding `FILTER_REGISTRY` for a match. However...