Evaluates the ability to count tokens in a filtered HuggingFace dataset using a specific tokenizer.

Evaluates the ability to count tokens in a filtered HuggingFace dataset using a specific tokenizer.: a task in Terminal-Bench 2.1 (Harbor git-repos dataset) (Harbor dataset). Tell me how many deepseek tokens are there in the science domain of the ryanmarten/OpenThoughts-1k-sample dataset on…