Hallucinated AI & agentic coding news. Some of it is real.

event

Dpo

46 items · started 2023-08-08 · ongoing (last activity 2026-09-16)

  1. Automated depression screening from clinical interviews requires attribution of utterances to the clinician or patient. We evaluate two datasets: DAIC-WOZ, where participant-only recordings require re-synthesizing both sides for controlled…

  2. Continued pretraining (CPT) with data augmentation such as paraphrasing can store inside a large language model (LLM) the knowledge of a small source corpus. The stored knowledge, however, is not always retrieved correctly.

  3. Direct preference optimization DPO is a promising offline approach for aligning large language models (LLMs) due to its simplicity, computational efficiency, and implicit modeling of human preferences. Interestingly, iterative extensions o…

  4. Ask a language model to respond "very excitedly," and its output is typically only mildly more energetic. We quantify this effect.

  5. Large language models (LLMs) have recently shown promise for historical entity linking, but preference optimization for this task is often formulated with only one negative candidate per training instance. This discards information from th…

  6. Reward hacking during reinforcement learning from verifiable rewards (RLVR) can induce reward seeking and broad misalignment in language models. Studying this misgeneralization is important for developing better threat models and counterme…

  7. As large language models (LLMs) are increasingly deployed in alignment-sensitive contexts, activation steering has emerged as a lightweight, inference-time alternative to fine-tuning methods (e.g., RLHF, DPO) for behavioral control. Howeve…

  8. Preference-based fine-tuning methods such as RLHF and DPO require substantial compute and large preference datasets. They also need direct access to the model parameters which are not provided by many state-of-the art models.

  9. Language models can ignore prompt evidence when it conflicts with memorized knowledge. Post-training can make models follow such evidence more reliably, but it is unclear whether these gains require new machinery or strengthen machinery al…

  10. Multimodal search agents answer visual questions by interleaving image understanding, web retrieval, tool use, and evidence synthesis. Strong systems exist, but in two expensive regimes: proprietary frontier models such as GPT-5 and Gemini…

  11. Direct Preference Optimization (DPO) simplifies alignment through pairwise comparisons but assumes all observed preferences are reliable. Real data often violates this assumption, leading to reversed, weak, or ambiguous labels that cause h…

  12. Arabic large language models must refuse harmful prompts without over-refusing benign or sensitive prompts, yet a single refusal rate hides this trade-off. We evaluate it using benign refusal B and harmful-prompt refusal H, where H measure…

  13. Direct Preference Optimization (DPO) is a widely used objective for aligning language models from preference data, with the coefficient $\beta$ commonly interpreted as controlling the KL constraint to a reference policy. We show that $\bet…

  14. Direct Preference Optimization (DPO) aligns Large Language Models with human preferences without the need for a separate reward model. However, DPO treats all tokens in responses equally, neglecting the differing importance of individual t…

  15. Free, gamified roadmaps for LLM engineering: an Inference Engineering path (KV caches, CUDA kernels, production vLLM serving) and a Model Training path (pretraining on a budget, scaling laws, SFT/DPO/GRPO) — 183 tasks with auto-verified mi…

  16. How precisely can we tell a language model how to feel? Most work on emotional generation answers with a discrete label - happy, angry, sad - which cannot express a target like "mildly downcast but calm." We instead specify the desired aff…

  17. Cyclic peptides are emerging as promising molecular scaffolds in drug discovery due to their high binding affinity and structural stability. However, extending generative models from linear to cyclic peptide design remains challenging, as…

  18. Direct Preference Optimization (DPO) has emerged as an effective approach for aligning large language models (LLMs) with human preferences. However, its adaptation to multimodal settings remains unexplored.

  19. Data selection in supervised fine-tuning aims to select a small set of effective samples from large-scale candidate data, reducing training cost while preserving model performance. However, existing methods usually treat data value as a re…

  20. Aligning large language models with human preferences must balance two competing goals: responding helpfully to legitimate requests and reliably refusing harmful ones. Most preference-based safety alignment methods collapse safety into a s…

  21. Achieving effective self-correction, where models verify and correct their own mistakes, remains a fundamental challenge for large language models (LLMs). In this work, we propose Self-Fix Step-DPO (SFS-DPO), a reinforcement learning based…

  22. The dominant paradigm treats AI safety as a property to be instilled during model training via RLHF, DPO, or Constitutional AI. We argue this is structurally insufficient for autonomous agents that execute code, mutate files, send messages…

  23. Direct Preference Optimization (DPO) aggregates token-level log-probability ratios via uniform summation, implicitly treating all tokens as contributing equally to the preference signal. However, the contribution of individual tokens to th…

  24. Minimal LLM Post-Training on an 8GB GPU: Understanding KL, SFT, DPO, GRPO and DeepSeek-Style Reasoning with Open-Source Frameworks Using open-source training frameworks (HuggingFace TRL) and minimal, reproducible experiments to see — one b…

  25. About Press Copyright Contact us Creators Advertise Developers Terms Privacy Policy & Safety How YouTube works Test new features NFL Sunday Ticket © 2026 Google LLC

  26. Recent Text-to-SQL methods rely heavily on reasoning-centric paradigms such as Chain-of-Thought (CoT), achieving substantial gains on complex benchmarks at the cost of high inference-time overhead. However, a large fraction of real-world q…

  27. The sycophancy of large language models can increase the safety risk in intervention dialogue for autistic children. Supervised fine-tuning can somewhat reduce sycophancy, but relying solely on positive examples is often insufficient to id…

  28. Direct Preference Optimization (DPO), the efficient alternative to PPO-based RLHF, falls short on knowledge-intensive generation: standard preference signals from human annotators or LLM judges exhibit a systematic verbosity bias that rewa…

  29. Large Language Model (LLM) agents are commonly trained from expert trajectories using supervised fine-tuning (SFT), which treats multi-turn agent behavior as ordinary text imitation. This recipe is simple and low-cost, but it only learns t…

  30. Direct Preference Optimization (DPO) is an effective framework for aligning large language models with human preferences, but it struggles with complex reasoning tasks. DPO optimizes for the likelihood of generating preferred over disprefe…

  31. RLHF evaluation should track how failures emerge, where they localize, and which warning signals appear before external quality degrades. We study this problem with a compact RLHF pipeline built for this paper, including PPO, DPO, uncertai…

  32. Long-context dialogue systems suffer from state inertia, where models over-attend to history and fail to adapt to evolving intents. We demonstrate that standard alignment methods like DPO and even recent long-context optimization technique…

  33. Normative theories allow one to elicit key parts of a ML algorithm from first principles, which is crucial at a time of championed scrutiny for ML work. Direct Preference Optimization (DPO) cleverly bypasses reward modeling by making an ex…

  34. Industrial LLM teams often ship behavior updates by repeatedly DPO-training a base model on sequences of related preference-data campaigns. The dominant failure mode in this regime is not always classical catastrophic forgetting: a pipelin…

  35. We present an approach to fine-tuning large language models using Direct Preference Optimization (DPO), a reinforcement learning technique. Our experimental results demonstrate that DPO simplifies the training pipeline, improves computatio…

  36. I have a budget of 5K, and want to buy some gpus my requirement is 48gb+ vram, because I finetune small language model, perform DPO, in general tinkering/ development is my usecase. if you where in my shoe which among these would you get,…

  37. Built a decision-reasoning engine (Orlog) and wanted to fine-tune a local model for it instead of paying per-call forever. The method (DV-DPO): Run a 3-voice council on each question, produce a synthesis Cross-examine: losing voices challe…

  40. When does autoresearch need a human? Autonomous research agents are everywhere in AI research workflows now.

  41. Probe-Detected Grokking in Multi-Probe DPO Orthogonal Learning Beyond Task-Specific Detectors in Qwen3.6-27B Probe-Detected Grokking in Multi-Probe DPO: Orthogonal Learning Beyond Task-Specific Detectors Abstract We report a phase-transiti…

  42. huggingface/ml-intern is HF's autonomous ML engineer — reads papers, audits datasets, ships SFT/DPO/LoRA/GRPO runs to HF Jobs. it's a standalone python harness with its own agent loop calling the Claude API.

  43. github link : genji970/hallucination-mitigation-via-contrastive-sampling-method: Selective contrastive post-training for hallucination mitigation in LLMs — improves factuality with ~10% data. ## Experimental Results ### (a) DPO vs.

  45. Hey folks, this is for those who appreciate experimentation on open-sourced AI models. We fine-tuned open-sourced SMLs (3B and 7B parameters) with SFT + DPO against commercial models like GPT-5.4, Gemini 3.1 Pro, Claude Opus 4.6, Google Do…

← all threads