clm-projects/nemotron-image-training-v3-qwen3.6-35b-a3b-fp8-embeddings

Nemotron-Image-Training-v3 · Qwen3.6-35B-A3B-FP8 embeddings State/action embedding pairs for CLM image pre-training, in the clm_train embedding-dir layout ({split}/state_embeddings.pt, {split}/action_embeddings.pt, {split}/metadata.json), the same layout as the image-v3.5-qwen3.6 store. Source: nvidia/Nemotron-Image-Training-v3 (CC BY 4.0) at revision 7656391, the nine subsets that ship their media. Encoder: Qwen3.6-35B-A3B-FP8 through vLLM, LAST-token pooling, hidden size 2048… See the full description on the dataset page: https://huggingface.co/datasets/clm-projects/nemotron-image-training-v

<iframe src="https://silytics.si/embed/item/6e026055-4cae-4335-a1ce-a888d119fc6d" width="420" height="190" style="border:0" loading="lazy" title="Silytics entry"></iframe>

Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 23 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.