Research

• Audio and music generation · Model · ICLR 2026

  TangoFlux

  Text-to-audio generation with flow matching and CRPO preference optimisation. Generates 30 seconds of 44.1 kHz audio in a few seconds on a single A40.

  886 stars 877/mo

  Code Weights Paper Demo

• Audio and music generation · Model · ACM MM 2023

  Tango

  An open latent-diffusion model for text-to-audio, built on an instruction-tuned LLM text encoder rather than a contrastive one.

  1,240 stars 67/mo

  Code Weights Paper

• Audio and music generation · Model · NAACL 2024

  Mustango

  Text-to-music generation conditioned on explicit musical structure — chords, beats, tempo and key — rather than free-text mood alone.

  396 stars 1,778/mo

  Code Weights Paper

• Audio and music generation · Model · 2025

  JAM

  A compact flow-based song generator with word-level lyric timing control and aesthetic alignment from synthetic preference pairs.

  170 stars 100/mo

  Code Weights Project

• Audio and music generation · Dataset · 2023

  TangoPromptBank

  The text-audio caption corpus Tango was pretrained on.

  320/mo

  Data

• Embodied AI and robotics · Model · 2025

  NORA

  A 3B-parameter open generalist vision-language-action model trained on 970k real-world robot demonstrations, built for low inference cost.

  224 stars 611/mo

  Code Weights Paper Project

• Embodied AI and robotics · Model · 2025

  NORA-1.5

  NORA with a flow-matching action expert, post-trained against world-model and action-based preference rewards.

  109 stars 38/mo

  Code Weights Overview Full demo

• Embodied AI and robotics · Model · 2025

  Emma-X

  An embodied multimodal action model that plans with grounded chain-of-thought and look-ahead spatial reasoning.

  84 stars 80/mo

  Code Weights Paper Data

• Trustworthiness and safety · Model + data · ICLR 2025

  Trust-Align

  Measures whether RAG answers are supported and fully cited, and trains models to refuse when the retrieved evidence is insufficient.

  76 stars 196/mo

  Code Data Paper

• Trustworthiness and safety · Code · ACL 2024

  RESTA

  Restoring safety in fine-tuned language models by adding a safety vector back through task arithmetic, without losing downstream capability.

  33 stars

  Code Paper

• Trustworthiness and safety · Code + data · 2023

  Red-Instruct

  Chain-of-Utterances red-teaming, with the RED-EVAL benchmark and the HarmfulQA dataset for probing and repairing harmful behaviour.

  111 stars 1,943/mo

  Code Data Paper

• Trustworthiness and safety · Benchmark · ICLR 2026

  OffTopicEval

  Measures whether a task-scoped LLM agent accepts what falls inside its remit and refuses what falls outside it.

  12 stars 130/mo

  Code Data Paper

• Efficiency, memory and agents · Code · ICML 2026

  Data Agent

  Learns which training examples to select as the target model changes, using difficulty and uncertainty as feedback.

  14 stars

  Code Paper Lab note

• Efficiency, memory and agents · Model · 2026

  δ-mem

  Online associative memory for LLMs — a 0.12% parameter addition that modulates attention through a low-rank interface while the backbone stays frozen.

  261 stars

  Code Paper Lab note

• Efficiency, memory and agents · Code · 2026

  IDEAgent

  Searches for a set of research ideas that meet quality thresholds without repeating one another.

  19 stars

  Code Paper Lab note

• Efficiency, memory and agents · Code · 2026

  EFLA

  Error-free linear attention derived as an exact solution from continuous-time dynamics rather than an approximation.

  77 stars

  Code

• Efficiency, memory and agents · Model · 2023

  Flan-Alpaca

  Extending Stanford Alpaca's synthetic instruction tuning to Flan-T5, with the Flan-mini instruction collection released alongside.

  354 stars 2,065/mo

  Code Weights Data

• Efficiency, memory and agents · Benchmark · 2023

  InstructEval

  Evaluates instruction-tuned models on problem solving, writing and alignment to human values.

  553 stars 48/mo

  Code Data Paper

• Multimodal understanding · Dataset · ACL 2019

  MELD

  Multimodal multi-party emotion recognition in conversation, used in later audio and multimodal evaluation work.

  1,088 stars 2,082/mo

  Code Data Paper

• Multimodal understanding · Code · 2019

  conv-emotion

  Reference implementations of emotion-recognition-in-conversation models, including DialogueRNN.

  1,527 stars

  Code

• Multimodal understanding · Code · 2020

  multimodal-deep-learning

  A collection of multimodal representation-learning and fusion models for sentiment and emotion, kept as a single reference implementation set.

  919 stars

  Code

• Multimodal understanding · Code · ACM MM 2020

  MISA

  Decomposes multimodal representations into modality-invariant and modality-specific subspaces before fusion.

  296 stars

  Code Paper

• Multimodal understanding · Dataset · 2021

  RECCON

  Recognising the cause of an emotion in conversation, not only the emotion itself.

  191 stars

  Code

• Multimodal understanding · Dataset · 2022

  CICERO

  Dialogue-level commonsense inference — causes, motivations, reactions and consequences drawn from conversation.

  63 stars 175/mo

  Code Data

• Multimodal understanding · Benchmark · 2024

  PuzzleVQA

  Abstract visual puzzles and algorithmic reasoning tasks that separate perception from inductive reasoning in multimodal models.

  119 stars 427/mo

  Code Data AlgoPuzzleVQA

• Reading lists · Reading list · 2018

  Awesome Sentiment Analysis

  538 stars

  Code

• Reading lists · Reading list · 2019

  Awesome ERC

  284 stars

  Code

No releases match that search.

Updated 28 September 2026 · All 81 public GitHub repositories · 51 models and 21 datasets on Hugging Face