REVIEW 10 cited by
DistiLLM-2: A Contrastive Approach Boosts the Distillation of LLMs
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
Signed reviews
read the original abstract
Despite the success of distillation in large language models (LLMs), most prior work applies identical loss functions to both teacher- and student-generated data. These strategies overlook the synergy between loss formulations and data types, leading to a suboptimal performance boost in student models. To address this, we propose DistiLLM-2, a contrastive approach that simultaneously increases the likelihood of teacher responses and decreases that of student responses by harnessing this synergy. Our extensive experiments show that DistiLLM-2 not only builds high-performing student models across a wide range of tasks, including instruction-following and code generation, but also supports diverse applications, such as preference alignment and vision-language extensions. These findings highlight the potential of a contrastive approach to enhance the efficacy of LLM distillation by effectively aligning teacher and student models across varied data types.
Forward citations
Cited by 10 Pith papers
-
Poly-OPD: Heterogeneous Multi-Teacher On-Policy Distillation for Capability-Selectable Flow Models
Poly-OPD distills two heterogeneous text-to-image teachers into one 2.5B student via a pixel bridge and DINOv2-space supervision, surpassing both teachers on GenEval.
-
Cross-Tokenizer On-Policy Distillation via Byte-Prefix Marginalization
Byte-Prefix Marginalization maps a teacher's next-token distribution onto the student's vocabulary through shared byte prefixes plus an explicit residual, giving a mass-preserving target for on-policy distillation acr...
-
When Top-K Misses the Decision: Tool-Call Drift in Multi-Teacher On-Policy Distillation
Top-K teacher logits can preserve almost all probability mass while dropping the low-probability '<tool call>' token, biasing distilled models to over-use tools; restoring the omitted token reduces over-calling 14.2%→...
-
Weak-to-Strong Generalization via Direct On-Policy Distillation
Transferring the log-ratio of a small model's pre-RL and post-RL checkpoints provides a dense implicit reward that improves stronger student models at a fraction of the cost of direct RL.
-
A Unified Approach to Interpreting Knowledge Distillation for Large Language Models via Interactions
KD sparsifies complex interactions in student LLMs; a Complex Interaction Penalty that enforces this improves multiple KD methods on in- and out-of-domain benchmarks.
-
Any-OPD: Heterogeneous On-Policy Distillation for Flow-Matching Models via Representation-Space Bridging
A distillation method that supervises a student on its own samples using teacher refinements compared in a frozen DINOv2 feature space, reporting preference-metric gains on a FLUX.1-dev to SD3.5-Medium transfer.
-
Compete Then Collaborate: Frontier AI Teachers Build a Verifiable Curriculum to Improve a Coding Student Beyond Imitation
Imitation of multi-teacher verified coding solutions degrades a competent student, while RL with verifiable rewards on the same curriculum improves it.
-
Focusing by Contrastive Attention: Enhancing VLMs' Visual Reasoning
CARVE contrasts attention maps from a general prompt and a specific question to mask out visual noise, then re-asks the question on the cropped and enlarged image, improving VQA accuracy by up to 75% on some benchmarks.
-
Enhancing Reasoning Capabilities in SLMs with Reward Guided Dataset Distillation
AdvDistill uses group relative advantages computed from rule-based rewards to weight teacher responses during distillation, reportedly improving a 1.5B student on math tasks beyond its 7B teacher.
-
Being Strong Progressively! Enhancing Knowledge Distillation of Large Language Models through a Curriculum Learning Framework
POCL wraps LLM knowledge distillation in a curriculum that increases data difficulty and temperature over stages, improving Rouge-L on small GPT-2 and OPT students, though ablations show temperature drives the gains.
Discussion (0). Continue with ORCID to comment.