Open Models Almanac
Methodstraining objectiveauxiliary loss

methods · training objective

auxiliary loss

26 methods filed at this node or below it, from the sources of 19 models.

In this branch 26

Everything filed at this node or below it, with one collapsible heading per child node.

filed here 26

KL alignment loss used · 3 sources · 4 quotes
Knowledge insulation used · 3 sources · 5 quotes
Sample-level attention masking used · 2 sources · 2 quotes
22 single mentions
Auxiliary object-detection loss core · 1 source · 1 quote
Log-probability-ranked sentence masking core · 1 source · 1 quote
Natural masks for missing depth regions core · 1 source · 1 quote
Squared log-policy-ratio regularizer core · 1 source · 1 quote
Domain-specific KL regularization strength used · 1 source · 2 quotes
Gradient detach used · 1 source · 2 quotes
KL regularization to a reference policy used · 1 source · 1 quote
Language modeling loss plus KL loss used · 1 source · 1 quote
Loss masking of flagged trajectory content used · 1 source · 1 quote
MoE sequence auxiliary loss used · 1 source · 1 quote
Unfinished-trajectory loss masking used · 1 source · 1 quote

By model

Which of this branch's methods each model's own documents describe, and how strongly. Under each model: its strongest treatment anywhere in the branch.

Modelmethods
MiMo-V2.6-Flash usedLoss masking of flagged trajectory content used—
NVIDIA-Nemotron-3-Ultra-550B-A55B usedUnfinished-trajectory loss masking used—
DeepSeek-V4-Flash usedSample-level attention masking used—
Alpamayo 1.5 / 2 Super usedKL regularization to a reference policy usedKnowledge insulation used—
DeepSeek-V3.2 usedDomain-specific KL regularization strength usedKL alignment loss used—
DeepSeek-V4-Pro usedSample-level attention masking used—
GR00T N1.7 coreAuxiliary object-detection loss coreSquared-error loss for normalized object-center coordinates core—
Kolibri-1 coreLog-probability-ranked sentence masking corePer-step policy-dependent recomputation of context masks coreSquared log-policy-ratio regularizer core—
LFM3-1.7B usedWeighted averaging of audio codebook losses used—
LingBot-VLA 2.0 coreNatural masks for missing depth regions core—
MiMo-V2.5 usedMoE sequence auxiliary loss used—
MiMo-V2.6-Pro usedLoss masking of flagged trajectory content used—
MiniMax-M3 usedGradient detach usedKL alignment loss usedLanguage modeling loss plus KL loss used—
Mistral-Large-4 optionalReference-policy KL penalty using the k2 estimator optional—
MolmoAct 2 usedAttention-mask isolation of packed examples usedDepth-prefix token replacement noise during fine-tuning usedKnowledge insulation used—
NVIDIA-Nemotron-3.5-Lightning-30B-A3B usedDisable auxiliary and hard-target losses during QAD usedMask non-target tokens in rollout training data used—
NVIDIA-Nemotron-3.5-Super-VL-120B-A12B usedDisable auxiliary and hard-target losses during QAD usedMask non-target tokens in rollout training data used—
Xiaomi-Robotics-1 usedAuxiliary action-generation supervision for the vision-language model usedBeta-distributed flow-matching timestep sampling usedLoss masking for missing embodiment-specific action dimensions used—
π0.5 usedAttention masking to prevent information leakage between action representations usedKnowledge insulation used—
↑