AI·Virginia Tech researchers found that resistance training (modeled as mouse weightlifting) outperforms running for reducing fat and improving insulin sensitivity and glucose control in obesity and Type 2 diabetes models.
Search
What are you looking for?
15 results for “ml”
AI·Achieves 50-54 tok/s for Qwen3 30B A3B float8 on 16GB RTX 5060 Ti using custom CUDA/C++.
AI·Researchers developed SQZ-Qwen-ASR-1.7B for MLC-SLM 2026 Task 1, combining a modular speaker diarization front-end (VAD, CAMPPlus embeddings, spectral clustering, RTTM segmentation) with Qwen3-ASR-1.7B adapted via full supervised fine-tuning, LoRA on TTS-generated synthetic speech, and GRPO reinforcement learning for lower tcpMER (23.70 on dev set, 17.97 on eval set).
AI·Readers are pleasantly surprised to discover O'Reilly books on machine learning available at a public library.
AI·Qwen3.6 35B-A3B Q8_0 model (no KV quant) generates full HTML flight simulator in single prompt and is praised as stronger than Q4_K_M on GPU due to better CPU performance.
I am currently working across multiple research communities, and I've noticed that the ML community is struggling with a massive volume of submissions, which is affecting review quality (as we are seeing in the recent ARR cycles). I am wondering what the reasoning is for not limiting the number of submissions per author? This practice has been successfully used in other research areas for years, …
So, I am working on this startup project with pretty low budget and one of the features is sentiment analysis based on political news, x posts and Instagram hashtag trends in which will be in Indian languages. I've been suggested muRIL, an Indian language-based model fine-tuned on political data as the best long-term option. But our team does not have any ML engineer so we dont know how we should…
What it is Talos-XII is a CLI simulator for the gacha system in Arknights: Endfield. Rather than sampling from a static probability table, it trains a small set of neural nets to model environment uncertainty and pull-decision policy, then uses them to answer questions a static table can’t easily express — e.g. “as a F2P player, what’s my probability of getting the rate-up unit on free currency a…
Lately in the last two/three years, I have noticed ICML, Neurips becoming more prestigious than the actual journals. What is the actual reason of this culture? Is this due to the AI boom and rising demand and the fact that conferences have a higher and a faster acceptance rate as compared to journals and with the growing hype they need to deliver things faster? What do you all think? submitted by…
Accurate breast cancer classification from mammography requires effective integration of complementary information from craniocaudal (CC) and mediolateral oblique (MLO) views, which provide a more complete characterization of breast abnormalities. However, existing multi-view learning approaches typically rely on feature-level aggregation or single-stage cross-attention, which can entangle view-specific and shared representations and restrict interaction to limited network depths. To address th…
We introduce PAST-TIDE, our stance detection system addressing both subtasks of the StanceNakba Shared Task at NakbaNLP@LREC-COLING 2026. The main idea is statement tuning. We redefine stance as cloze-style masked language modeling (MLM), letting a verbalizer map label words to stance categories through the pre-trained MLM head rather than appending a randomly initialized classification head. We complement this with prototypical contrastive learning, which uses learnable class prototypes for ba…
Multimodal large language models (MLLMs) generate responses autoregressively, integrating visual and linguistic information in an evolving context. Prior work on interpretability has focused on individual layers and circuits (where), leaving the token-level dynamics of multimodal computation during generation (when) underexplored. We address this gap and study attention shifts as per semantic role; tracking model attention to image, text, instruction, and previously generated tokens, One Token …
Reinforcement learning (RL) has become a central component of post-training large language models (LLMs), yet little is understood about how RL adaptation is distributed across transformer layers. Existing approaches typically update all model parameters uniformly, implicitly assuming that every layer contributes similarly to the gains obtained during RL post-training. In this work, we challenge this assumption through a systematic layer-wise study of RL training. Surprisingly, we find that tra…
Touch supplies the physical grounding needed to perceive intrinsic material properties, such as friction and compliance, that vision alone often cannot resolve. Recent efforts for equipping multimodal LLMs with this tactile sense, however, expose a zero-sum trade-off: the limited parameter budget of compact models forces a choice between acquiring the new sensory modality and preserving the established vision-language reasoning. We present Splash, a mask-isolated tactile alignment learning fram…
Hi everyone, I recently finished my Ph.D. thesis on Differentiable Ray Tracing for Radio Propagation Modeling. Instead of just compiling my published papers, I tried to write it as an accessible, self-contained textbook for anyone interested in the intersection of radio propagation simulation, autodiff, and ML. Permanent handle: https://hdl.handle.net/2078.5/278727 Repo with TeX source files Whil…