arXiv AI
Jun 9
Phantom transitions in language model fine-tuning
arXiv:2606. 07559v1 Announce Type: cross Abstract: Fine-tuning a language model on contexts whose correct completion has a near-synonym competitor often fails silently.
Phantom Transitions in Language Model Fine-Tuning: A Density-Matrix Analysis
arXiv:2606. 07559v2 Announce Type: replace-cross Abstract: Fine-tuning a language model often fails silently when its correct completion must outrank a near-synonym competitor.
Related stories
arXiv Machine Learning
Aug 12
Off-Axis, On Purpose: Where a Transformer Computes Concepts and Why it Does So
arXiv:2608. 10251v1 Announce Type: cross Abstract: A transformer's answer lives on one axis: the direction its unembedding reads.
arXiv Machine Learning
Aug 7
Cross-Architecture Steering Transfer in Language Models: A Systematic Empirical Study
arXiv:2608. 05164v1 Announce Type: cross Abstract: Independently trained large language models may develop shared internal representations of semantic concepts despite architectural differences -- but whether this geometric similarity has functional consequences for cross-model behavioural control remains untested.
arXiv Machine Learning
Sep 3
The Dynamics of Continuous Mixture Collapse in Language Models
The paper investigates why large language models (LLMs) fail to maintain continuous mixtures of token embeddings—used in latent-state reasoning—to preserve multiple reasoning paths. Through theory and experiments, it identifies three failure sources: transformer geometry distortion, amplification or contraction dynamics from softmax and autoregressive feedback, and the need for context-dependent corrections that scale with mixture size. Empirical results confirm the predicted transition between contraction and amplification and show pretrained models largely fall on the amplifying side.
arXiv AI
Sep 15
Ventriloquist LLMs: Linear Alignment of Late-Stage Representations
arXiv:2603.18908v5 Announce Type: replace Abstract: Independently trained language models often learn compatible late-stage representations, despite differences in training objectives, architectures,...
arXiv AI
Jul 8
Transformers converge to invariant algorithmic cores
arXiv:2602. 22600v2 Announce Type: replace-cross Abstract: Training selects for behavior, not circuitry: many weight configurations can implement the same function.
Hugging Face Trending Papers
Jun 25
Structure Before Collapse: Transient semantic geometry in next-token prediction
Neural Collapse predicts that balanced one-hot classification pushes model representations to be equally far from each other; a symmetric configuration that depends only on the output label and ignores any semantic similarity in the inputs. This creates a puzzle: next-token prediction language models are trained predominantly (as context length increases) with one-hot labels: the same context is very unlikely to appear twice in training with different labels.
arXiv Machine Learning
4d ago
The Geometry of Inference in Transformer Residual Streams
arXiv:2609.37824v1 Announce Type: cross Abstract: Transformer language models build predictions through successive residual updates, but how their representations become specific to an eventual outco...
arXiv AI
2d ago
How Divergence Becomes Decision Flips in Compressed Language Models
arXiv:2610.00694v1 Announce Type: cross Abstract: Compression reports summarize how far a compressed language model moved from the dense one, usually by a KL divergence; a deployment that relies on t...
arXiv Computation and Language
Sep 1
Where Do Multilingual Vision-Language Encoders Fail on Low-Resource Languages?
arXiv:2608.30725v1 Announce Type: new Abstract: Recent multilingual vision--language encoders cover hundreds of languages in a single model, yet on two state-of-the-art instances retrieval on low-res...
arXiv Machine Learning
Aug 19
FishBack: Pullback Fisher Geometry for Optimal Activation Steering in Transformers
FishBack introduces a pullback Fisher geometry approach for activation steering in transformers, challenging the common Euclidean assumption of intermediate activation spaces. By deriving a closed‑form steering direction based on the Fisher information metric of the softmax layer, the method achieves target concept changes with minimal off‑target distortion, especially in early and middle layers. Experiments on GPT‑2 Small, Llama‑3‑8B, and Qwen3‑8B demonstrate significant reductions in off‑target KL divergence compared to existing steering baselines.
arXiv AI
3d ago
Does This Action Still Explain the Task? Reverse Scoring for Diffusion Language Model Agents
arXiv:2609.38536v1 Announce Type: cross Abstract: Diffusion-based large language models (dLLMs) promise to break the sequential latency bottleneck of autoregressive agents through parallel decoding,...