菜单
Reduce hallucinations in multimodal reasoning by detecting high-entropy (uncertain) states and switching to continuous latent embeddings instead of discrete tokens. Use prior-guided visual anchoring during uncertain phases to maintain grounding.
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/ADu2021/skillXiv --skill latent-entropy-aware-decoding命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
Convert arXiv papers into ready-to-use agent skills using category-aware extraction. First classifies the paper into one or more of 11 research categories, then applies a specialized extraction pipeline for each category — because different types of papers produce different types of usable knowledge. A single paper can yield multiple skills if it spans categories. Use this skill whenever the user wants to turn a paper into a skill, extract practical techniques from research, build a skill library from papers, convert arXiv papers into reusable agent instructions, or batch-process multiple papers into skills. Also trigger when someone asks about extracting actionable knowledge from papers, making research practical for LLM agents, or systematically converting academic contributions into structured agent capabilities.
Establish regret bounds for behavior cloning with discretized actions combining statistical error and quantization error terms. Prove smoothness requirements for safe quantizer design, show that learning-based quantizers fail these requirements, and propose model-based augmentation to reduce error dependence from H² to H.
Dynamically allocate LoRA ranks per-layer during fine-tuning instead of using fixed uniform ranks. Learn optimal rank for each layer and subject via variational framework with discretized exponential distribution, reducing memory footprint while maintaining fidelity and text-alignment.
基于 SOC 职业分类
正在显示 SKILL.md
| name | latent-entropy-aware-decoding |
| title | Thinking in Uncertainty: Mitigating Hallucinations in MLRMs with Latent Entropy-Aware Decoding |
| version | 0.0.2 |
| engine | skillxiv-v0.0.2-claude-opus-4.6 |
| license | MIT |
| url | https://arxiv.org/abs/2603.13366 |
| keywords | ["Hallucination Mitigation","Entropy-Aware Decoding","Multimodal Reasoning","Uncertainty Quantification","Continuous Embeddings"] |
| description | Reduce hallucinations in multimodal reasoning by detecting high-entropy (uncertain) states and switching to continuous latent embeddings instead of discrete tokens. Use prior-guided visual anchoring during uncertain phases to maintain grounding. |
Multimodal large reasoning models (MLRMs) frequently hallucinate during high-uncertainty reasoning phases, particularly at transition points where the model is uncertain about causal structure (words like "because," "however," "wait"). LEAD (Latent Entropy-Aware Decoding) mitigates this by monitoring the entropy of the model's latent representations. During high-uncertainty phases, rather than committing to discrete token selections, the model maintains superposed probability-weighted embeddings capturing multiple competing hypotheses. This allows the model to extract contextual information more reliably and reduces unfounded claims during uncertain reasoning.
The technique is plug-and-play and works with existing MLRMs by modifying the decoding strategy, not the underlying architecture.
LEAD operates through entropy-aware mode switching:
The key insight: at high-entropy states, the model is uncertain about which discrete token to emit, but the continuous probability distribution contains useful information. By leveraging that distribution, you avoid premature commitment to hallucinated tokens.
Begin by computing entropy over model logits and detecting uncertainty phases during generation.
import torch
import torch.nn.functional as F
numpy np
:
():
.entropy_threshold = entropy_threshold
.entropy_history = []
() -> torch.Tensor:
probs = F.softmax(logits, dim=-)
entropy = -(probs * torch.log(probs + )).(dim=-)
entropy
() -> :
entropy = .compute_entropy(logits)
is_high_entropy = entropy > .entropy_threshold
.entropy_history.append(entropy.item())
is_high_entropy.item()
() -> :
uncertainties = [.is_uncertain(logits) logits logits_sequence]
regions = []
start =
i, is_unc (uncertainties):
is_unc start :
start = i
is_unc start :
regions.append((start, i))
start =
start :
regions.append((start, (uncertainties)))
regions
Next, implement continuous embedding generation during high-entropy states.
class ContinuousEmbeddingDecoder:
"""Generate superposed embeddings instead of discrete tokens during uncertainty."""
def __init__(self, embedding_matrix: torch.Tensor, temperature=0.7):
self.embedding_matrix = embedding_matrix # [vocab_size, embed_dim]
self.temperature = temperature
def discrete_decode(self, logits: torch.Tensor) -> int:
"""Standard: select token with highest probability."""
token_id = torch.argmax(logits, dim=-1)
return token_id.item()
def continuous_decode(self, logits: torch.Tensor) -> torch.Tensor:
"""Uncertain: create superposed embedding from probability distribution."""
# Temperature-scaled probabilities
scaled_logits = logits / self.temperature
probs = F.softmax(scaled_logits, dim=-1)
# Weighted combination of embeddings
# probs: [vocab_size], embedding_matrix: [vocab_size, embed_dim]
continuous_embedding = torch.matmul(probs, self.embedding_matrix)
# [embed_dim]
return continuous_embedding
def hybrid_decode(self, logits: torch.Tensor, entropy: float,
threshold: float = 1.5) -> torch.Tensor:
"""Choose decoding based on entropy."""
if entropy > threshold:
return self.continuous_decode(logits)
else:
# For discrete, return embedding of selected token
token_id = self.discrete_decode(logits)
return self.embedding_matrix[token_id]
Now implement visual grounding that reinforces factual anchoring during uncertain phases.
class VisualAnchor:
"""Inject visual information to ground reasoning during uncertainty."""
def __init__(self, visual_embed_dim=768, text_embed_dim=768):
self.visual_embed_dim = visual_embed_dim
self.text_embed_dim = text_embed_dim
# Alignment projection
self.text_to_visual = torch.nn.Linear(text_embed_dim, visual_embed_dim)
self.visual_to_text = torch.nn.Linear(visual_embed_dim, text_embed_dim)
def compute_visual_prior(self, visual_embeddings: torch.Tensor,
question_embedding: torch.Tensor) -> torch.Tensor:
"""Compute prior distribution over visual regions relevant to question."""
# visual_embeddings: [num_regions, visual_embed_dim]
# question_embedding: [text_embed_dim]
# Project question to visual space
question_in_visual = self.text_to_visual(question_embedding)
# Compute relevance scores
relevance = torch.matmul(visual_embeddings, question_in_visual)
relevance = F.softmax(relevance, dim=0)
return relevance
def inject_visual_anchor(self, text_embedding: torch.Tensor,
visual_embeddings: torch.Tensor,
visual_prior: torch.Tensor,
injection_strength: float = 0.3) -> torch.Tensor:
"""Blend text embedding with weighted visual information."""
# Aggregate visual embeddings using prior as weights
aggregated_visual = torch.matmul(visual_prior, visual_embeddings)
# Project to text space
visual_in_text = self.visual_to_text(aggregated_visual)
# Blend: higher injection_strength = more visual influence
anchored = (1.0 - injection_strength) * text_embedding + \
injection_strength * visual_in_text
return anchored
Finally, integrate entropy-aware decoding into the generation loop.
class LatentEntropyAwareDecoder:
"""Full decoding pipeline with entropy-aware mode switching."""
def __init__(self, model, tokenizer, embedding_matrix, visual_model=None):
self.model = model
self.tokenizer = tokenizer
self.entropy_monitor = EntropyMonitor(entropy_threshold=1.5)
self.continuous_decoder = ContinuousEmbeddingDecoder(embedding_matrix)
self.visual_anchor = VisualAnchor() if visual_model else None
self.visual_model = visual_model
def generate(self, prompt: str, visual_input=None, max_length=256):
"""Generate with entropy-aware mode switching."""
token_ids = self.tokenizer.encode(prompt)
generated = []
embeddings_history = []
for step in range(max_length):
# Get model logits
with torch.no_grad():
outputs = self.model(torch.tensor([token_ids]))
logits = outputs.logits[0, -1, :] # Last token logits
# Compute entropy
entropy = self.entropy_monitor.compute_entropy(logits)
# Choose decoding strategy based on entropy
if entropy > self.entropy_monitor.entropy_threshold:
# High entropy: use continuous embedding
embedding = self.continuous_decoder.continuous_decode(logits)
# Apply visual grounding if available
if self.visual_anchor and visual_input is not None:
question_embedding = self.model.encode_text(prompt)
visual_embeddings = self.visual_model(visual_input)
visual_prior = self.visual_anchor.compute_visual_prior(
visual_embeddings, question_embedding)
embedding = self.visual_anchor.inject_visual_anchor(
embedding, visual_embeddings, visual_prior,
injection_strength=0.4)
embeddings_history.append(embedding)
# For output, select token closest to continuous embedding
distances = torch.norm(
self.continuous_decoder.embedding_matrix - embedding,
dim=1)
next_token = torch.argmin(distances).item()
else:
# Low entropy: standard discrete decoding
next_token = torch.argmax(logits, dim=-1).item()
embeddings_history.append(
self.continuous_decoder.embedding_matrix[next_token])
generated.append(next_token)
token_ids.append(next_token)
# Stopping criteria
if next_token == self.tokenizer.eos_token_id:
break
return self.tokenizer.decode(generated)
def evaluate_hallucination_mitigation(model, test_cases, visual_inputs=None):
"""Measure hallucination reduction with LEAD."""
decoder = LatentEntropyAwareDecoder(model, tokenizer, embedding_matrix,
visual_model=vision_model)
hallucination_scores = []
for i, test in enumerate(test_cases):
visual = visual_inputs[i] if visual_inputs else None
output = decoder.generate(test['prompt'], visual_input=visual)
# Score hallucination (factuality vs. reference)
hallucination_score = compute_factuality(output, test['reference'])
hallucination_scores.append(hallucination_score)
avg_factuality = np.mean(hallucination_scores)
print(f"Average Factuality Score: {avg_factuality:.3f}")
return hallucination_scores
Hyperparameters and When to Use:
When NOT to use:
Common Pitfalls: