Skip to content
#

decoding-strategies

Star

Here are 4 public repositories matching this topic...

Mmorgan-ML / Phase-Slip-Sampler

Star 7

Phase-Slip is a stochastic intervention architecture that operates on the Key-Value Cache of the model. Phase-Slip gently rotates the semantic vectors of the context window, asking the model: "How would you finish this sentence if you looked at it from a slightly different perspective?"

python nlp open-source machine-learning entropy deep-learning transformers inference text-generation pytorch artificial-intelligence thermodynamics sampling simulated-annealing hallucination huggingface kv-cache llm hallucination-mitigation decoding-strategies
  • Updated Dec 18, 2025
  • Python

unshDee / sampler-scope

Star 0

How much of an LLM agent’s apparent behavior comes from the decoder rather than the model alone? SamplerScope holds the model’s logits fixed, changes only the decoding method, and measures how this alters outcomes in small decision environments.

sampling policy-evaluation markov-decision-process huggingface-transformers large-language-models qwen llm-evaluation llm-agents inference-time ai-welfare decoding-strategies decoder-attribution
  • Updated Aug 17, 2026
  • Python

Add this topic to your repo

To associate your repository with the decoding-strategies topic, visit your repo's landing page and select "manage topics."

Learn more