Update org card: add motion/IMU (Tremor), k3-vision, cookbook, and Fusion Perception

README.md

@@ -9,73 +9,28 @@ pinned: false
 
 ![Eximius Labs](assets/banner.png)
 
-This is Eximius Labs' home for open model weights and multimodal embedding
-research. Everything here gives developers and researchers a production-ready
-starting point for cross-modal retrieval and semantic search, backed by the
-research behind the Fusion Embedding stack. The technical report is on
-arXiv: https://arxiv.org/abs/2607.18666.
+Eximius Labs builds open-weight multimodal models that run on your own hardware: the memory and perception layer for physical AI. The technical report is on arXiv: https://arxiv.org/abs/2607.18666.
 
-## The Fusion Embedding family
+We freeze a strong base model and add new senses from the outside, so existing embeddings stay valid and gaining a modality never forces a re-index. One shared vector space spans text, image, video, audio, thermal, and motion (IMU).
 
-One embedding space for text, images, video, audio — and, with its first
-sense pack, thermal — built on a single design constraint: the base model
-is never retrained. Both generations are
-live on the public MTEB leaderboards for audio (MAEB) and video (MVEB),
-evaluated through the official `mteb` harness anyone can rerun.
+## Fusion Embedding, the memory layer
 
-**[fusion-embedding-1-2b-preview](https://huggingface.co/EximiusLabs/fusion-embedding-1-2b-preview)** (v0.3) —
-generation 1 trains a 16.4M-parameter connector between a frozen
-Qwen2.5-Omni audio tower and a byte-frozen Qwen3-VL-Embedding-2B base:
-under 1% of the model. AudioCaps audio-to-text R@10 0.741, at the frontier
-of models that keep their base frozen. Aligned on audio-text only, it
-retrieves images from sound at 29x chance with zero paired audio-visual
-training data.
+One vector space for cross-modal retrieval, RAG, clustering, and semantic search.
 
-**[fusion-embedding-2-2b-preview](https://huggingface.co/EximiusLabs/fusion-embedding-2-2b-preview)** (v0.2) —
-generation 2 adds modality-gated adapters inside the frozen decoder layers
-(+44.2M parameters), opened only for audio inputs. For text, images, and
-video the adapters never execute and outputs are bit-for-bit identical to
-the base model, verified after every training run. The family's best
-text-to-audio results on every benchmark: AudioCaps R@10 0.785, Clotho
-zero-shot 0.485.
+- [fusion-embedding-2](https://huggingface.co/EximiusLabs/fusion-embedding-2-2b-preview) — text, image, video, and audio in one space (connector plus modality-gated adapters).
+- [fusion-embedding-1](https://huggingface.co/EximiusLabs/fusion-embedding-1-2b-preview) — the connector-only first generation.
+- [Ember](https://huggingface.co/EximiusLabs/fusion-embedding-2-ember) — thermal and infrared, as a loadable sense pack.
+- [Tremor](https://huggingface.co/EximiusLabs/fusion-embedding-2-tremor) — body-worn and robot motion (IMU and accelerometer), read as language, with a [Unitree-G1 head](https://huggingface.co/EximiusLabs/fusion-embedding-2-tremor-g1).
+- [fusion-embedding-2-k3-vision](https://huggingface.co/EximiusLabs/fusion-embedding-2-k3-vision) — Kimi K3's vision encoder, projected into the shared space.
 
-**[fusion-embedding-2-ember](https://huggingface.co/EximiusLabs/fusion-embedding-2-ember)** (v0.1) —
-Ember is the thermal sense for fusion-embedding-2 and the first *sense
-pack*: a separately loadable adapter pack that adds a modality while every
-existing vector stays valid. Text-to-thermal retrieval R@10 0.785 on a
-2,000-image gallery against a 0.224 frozen baseline. Sense packs are named
-for the physical trace their sensor reads; Ember reads heat.
+## Fusion Perception, the perception layer
 
-## How it works
+Dense scene understanding and geometric place recognition on a frozen vision backbone, projected into the Fusion Embedding space. Code: https://github.com/Eximius-Labs/fusion-perception.
 
-A perceiver-resampler connector maps frozen audio-tower features into the
-base's input space; gated bottleneck adapters add in-layer capacity per
-modality. Gates are mutually exclusive by input modality, so non-gated
-modalities execute the base's exact computation. A single forward pass
-yields nested Matryoshka embeddings truncatable to
-{2048, 1536, 1024, 512, 256, 128, 64} dimensions. Training runs in hours on a single
-GPU; connectors and packs ship as small artifacts with the frozen towers
-downloaded separately.
+## Build with it
 
-```python
-import mteb
-model = mteb.get_model("EximiusLabs/fusion-embedding-2-2b-preview")
-```
+- Cookbook: https://github.com/Eximius-Labs/cookbook
+- Code: https://github.com/Eximius-Labs
+- Paper: https://arxiv.org/abs/2607.18666
 
-## Training data
-
-The audio corpus is ~500K open audio-caption pairs (AudioCaps and
-AudioCaps 2.0, FSD50K, WavCaps/AudioSet-SL, a LAION-FreeSound subset);
-Ember trains on IR-TD real thermal imagery with restricted sources removed.
-Evaluation sets are excluded from training by clip id at ingestion, and
-every reported number traces to a result record in the open repository.
-
-## Status
-
-English-language research previews. Model weights are released under
-CC-BY-NC-4.0 (Ember under its own research license; see each model card).
-Sound-event retrieval is the strongest axis; speech and music are
-undertrained in the current corpus. We release early to invite research and
-feedback on unified multimodal embedding spaces.
-
-[eximiuslabs.com](https://eximiuslabs.com/) · [github.com/Eximius-Labs](https://github.com/Eximius-Labs)
+Everything is open weight and self-hostable. Core models are research preview; the sensor packs are CC-BY-NC-4.0.