technique

encoder-free early fusion

techniqueactiveencoder-free-early-fusion-f5a27a7e·1 events·first seen 24d ago

Aliases: encoder-free early fusion

Co-occurring entities

Thinking Machines GPT-Realtime-2 FD-bench Google TML-Interaction-Small MiniCPM-o 4.5 Alibaba Tinker Qwen3.5 Omni Big Bench Audio Gemini 3.1 Flash Live Preview Audio MultiChallenge flow-matching decoder Mixture of Experts OpenAI Mira Murati

More like this (12)

Robust Dual-Signal Fusion encoder-only language models Cross-Event Prompt Fusion confidence gating FlashbackCL: Mitigating Temporal Forgetting in Federated Learning Feature Auto-Encoder temporally ordered pre-training intra-frame entropy-guided sparsification Fusion360 Sparse Autoencoders FusionRS Fast Adaptive Semantic Entropy

Recent events (1)

7The Batch·24d ago·source ↗

Thinking Machines Lab Reveals TML-Interaction-Small: Real-Time Multimodal Interaction Model

Thinking Machines Lab (founded by Mira Murati) has announced TML-Interaction-Small, a 276B-parameter mixture-of-experts multimodal model that processes audio, video, and text concurrently using 200ms 'micro-turns' rather than waiting for conversational turns to complete. The architecture uses encoder-free early fusion, pairing a fast foreground interaction model with an asynchronous background reasoning model that shares context. On interactivity benchmarks (FD-bench V1/V1.5), it outperforms GPT-Realtime-2 and Gemini-3.1-flash-live-preview, though it trails GPT-Realtime-2 on intelligence benchmarks. A closed research preview is expected in coming months with wider release later in 2026.

Frontier Model Releases Inference Economics encoder-free early fusion Thinking Machines GPT-Realtime-2 +16 more