SigLIP 2 Base NaFlex (RLX staging)

Google SigLIP 2 NaFlex (variable aspect) checkpoint for RLX.

Field Value
Hub id eugenehp/siglip2-base-naflex
Kind Staging redistrib of an upstream checkpoint for RLX runners.
RLX crate rlx-siglip2
Upstream https://huggingface.co/google/siglip2-base-patch16-naflex

Quick start

hf download eugenehp/siglip2-base-naflex --local-dir .
cargo run -p rlx-siglip2 --release -- --model-dir . --image doc.png --labels "a document, a photo"

File highlights

  • model.safetensors (1.4 GiB)
  • siglip2-base-naflex.f16.gguf (715.7 MiB)
  • tokenizer.json (32.8 MiB)
  • tokenizer.model (4.0 MiB)
  • fixture/reference.json (3.8 MiB)
  • tokenizer_config.json (39.2 KiB)
  • special_tokens_map.json (636 B)
  • preprocessor_config.json (393 B)
  • config.json (329 B)

Do not use as the main runtime pack

These files may be present for historical / staging reasons but are not the supported RLX load path:

  • siglip2-base-naflex.f16.gguf

Community *.f16.gguf / LM-only Q4 packs are format wraps โ€” they are not drop-in replacements for the RLX primary file above.

Run with RLX

Clone rlx-models, place this repo under weights/vision/siglip2-base-naflex (or pass the path explicitly), then:

cargo run -p rlx-siglip2 --release -- --model-dir . --image doc.png --labels "a document, a photo"

License

Apache License 2.0 โ€” see LICENSE. Inherit upstream terms when redistributing.

Original weights and authorship: https://huggingface.co/google/siglip2-base-patch16-naflex

Redistrib note

This Hub repo exists so RLX recipes have a stable fetch target. When you only need the upstream checkpoint, prefer the Upstream link above.

Downloads last month
68
Safetensors
Model size
0.4B params
Tensor type
F32
ยท
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support