emu3
Datasets
All datasets matching “emu3”unified-vlm-steering-emu35
Emu3.5 (BAAI): activation steering sweeps
Steered text and image generations from Emu3.5 (BAAI), one of the unified vision-language models in the unified-vlm-steering project. Steering adds alpha * v_hat (the per-layer unit difference-of-means vector) to the residual stream at every layer of a layer config.
34B dense autoregressive model, 64 decoder layers (0-indexed). Images are 32x32 IBQ tokens (512 px), generated on BAAI's patched vLLM engine with request-id-keyed CFG. The… See the full description on the dataset page: https://huggingface.co/datasets/saintsauce/unified-vlm-steering-emu35.Emu3_natureemu3-mscoco-train2017-tokens
Emu3 Inference Tokens on MS COCO (Train Subset)
Description
This repository contains token sequences generated by
BAAI/Emu3-Gen model on the MS COCO training dataset.
Only the first 1200 training samples are included since the inference is rather time consuming (about 400 seconds per image with 720x720 pixels).
Contents
Token sequences only,
Each .pt file corresponds to one MSCOCO training sample, stored using a Dict[str, torch.Tensor] with "tokens" as the… See the full description on the dataset page: https://huggingface.co/datasets/vlgunsdaddy/emu3-mscoco-train2017-tokens.emu3_tokenized_rl_100kcommoncatalog-cc-by-crop-128-emu3-encoded-datasetcommoncatalog-cc-by-crop-256-emu3-encoded-dataset