3/ MambaByte – adapts Mamba SSM to learn directly from raw bytes; bytes lead to longer sequences which autoregressive Transformers will scale poorly on; reports huge benefits related to faster inference and even outperforms subword Transformers.
@dair_ai
-
WARM: Weighted Averaged Rewards Models Improve LLM Alignment
By
–
5/ WARM – introduces weighted averaged rewards models (WARM) that involve fine-tuning multiple rewards models and then averaging them in the weight space; improves efficiency while improving the quality and alignment of LLM predictions.
-
Diffuse to Choose: Diffusion-Based Image Inpainting Model
By
–
4/ Diffuse to Choose – a diffusion-based image-conditioned inpainting model to balance fast inference with high fidelity while enabling accurate semantic manipulations in a given scene content.https://t.co/KUJjxVfCHc
— DAIR.AI (@dair_ai) 28 janvier 20244/ Diffuse to Choose – a diffusion-based image-conditioned inpainting model to balance fast inference with high fidelity while enabling accurate semantic manipulations in a given scene content.
-
Depth Anything: Monocular Depth Estimation from Unlabeled Data
By
–
1/ Depth Anything – a monocular depth estimation solution that can deal with any images under any circumstance; proposes effective strategies to leverage the power of the large-scale unlabeled data (~62M) which helps to reduce generalization error.https://t.co/cOXWWextRV
— DAIR.AI (@dair_ai) 28 janvier 20241/ Depth Anything – a monocular depth estimation solution that can deal with any images under any circumstance; proposes effective strategies to leverage the power of the large-scale unlabeled data (~62M) which helps to reduce generalization error.
-
Top Machine Learning Papers Week January 22-28
By
–
The Top ML Papers of the Week (Jan 22 – Jan 28): – WARM
– Medusa
– AgentBoard
– MambaByte
– Knowledge Fusion of LLMs
– Resource-efficient LLMs & Multimodal Models
… -
MoE-Mamba: Scaling LLMs with State Space Models and Mixture of Experts
By
–
10/ MoE-Mamba – an approach to efficiently scale LLMs by combining state space models (SSMs) with Mixture of Experts (MoE); MoE-Mamba, outperforms both Mamba and Transformer-MoE.
-
Easy Training Data Enables Hard Task Generalization
By
–
9/ The Unreasonable Effectiveness of Easy Training Data for Hard Tasks – suggests that language models often generalize well from easy to hard data, i.e., easy-to-hard generalization.
-
Patchscopes: Framework for Explaining LLM Internal Representations
By
–
8/ Patchscopes – proposes a framework that leverages a model itself to explain its internal representations; it can be used to answer questions about an LLM’s computation and can even be used to fix latent multi-hop reasoning errors.
-
LLM Evaluation Methodologies: Taxonomy and Approaches
By
–
7/ Overview of LLMs for Evaluation – thoroughly surveys the methodologies and explores their strengths and limitations; provides a taxonomy of different approaches involving prompt engineering or calibrating open-source LLMs for evaluation.
-
ReFT: Enhancing LLM Reasoning Through Reinforced Fine-Tuning
By
–
6/ Reasoning with Reinforced Fine-Tuning – an approach, ReFT, to enhance the generalizability of LLMs for reasoning; it starts with applying SFT and then applies online RL for further refinement while automatically sampling reasoning paths to learn from.