crosscoder-tokenized-hhrf-qwen / dataset_info.txt
Kasmic's picture
Upload folder using huggingface_hub
198e70f verified
raw
history blame contribute delete
353 Bytes
CrossCoder Dataset Info
=======================
Number of examples: 160800
Max sequence length: 1024
Tokenizer: Qwen/Qwen1.5-7B
Tensor shapes:
- chosen_input_ids: torch.Size([160800, 1024])
- chosen_attention_mask: torch.Size([160800, 1024])
- rejected_input_ids: torch.Size([160800, 1024])
- rejected_attention_mask: torch.Size([160800, 1024])