simpo-loss

simpo-loss is a skill for Claude Code, Codex from cxcscmu/SkillLearnBench. It costs 22 tokens per session (334 once invoked), scanned A, original, MIT.

A formula and implementation guide for SimPO, a method for training language models from preferred and rejected answers without a separate reference model.

In plain words
What is it for?
Use it when implementing or checking SimPO preference-training code in an LLM project.
Why use it?
It explains how to calculate the training loss and rewards, including sigmoid and hinge variants.

Skill for Claude CodeCodex

Written for no agent in particular: nothing here depends on one.

Good fit Use it when implementing or checking SimPO preference-training code in an LLM…

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/cxcscmu/skilllearnbench/simpo-loss
Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Any agent
npx skills add cxcscmu/SkillLearnBench --skill simpo-loss
Clone the repo
git clone --depth 1 https://github.com/cxcscmu/SkillLearnBench

Made for: Claude Code, Codex.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for simpo-loss

README.md
[![agentmods](https://agentmods.dev/badge/skills/cxcscmu/skilllearnbench/simpo-loss.svg)](https://agentmods.dev/skills/cxcscmu/skilllearnbench/simpo-loss)
Your own site
<a href="https://agentmods.dev/skills/cxcscmu/skilllearnbench/simpo-loss"><img src="https://agentmods.dev/badge/skills/cxcscmu/skilllearnbench/simpo-loss.svg" alt="Measured on agentmods" height="20"></a>
Per session 22 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 334 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00022 $0.00334
Opus 5 $0.00011 $0.00167
Sonnet 5 $0.00004 $0.00067
Haiku 4.5 $0.00002 $0.00033

Measured 3d ago against content hash 744a29e01e5d, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-06, from the pricing page.

Security

Grade A, and why

simpo-loss scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 3d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

skills/b1-one-shot-claude-opus-4-6/nlp-paper-reproduction/simpo-loss/SKILL.md · 40 lines

What it actually says

SimPO Loss

Overview

SimPO is a reference-free preference optimization algorithm. Its key innovation is using the average log probability of a sequence as the implicit reward, plus a target reward margin γ.

Loss Formula (Eq. 6 from the paper)

L_SimPO = -E log σ(β/|yw| · log πθ(yw|x) - β/|yl| · log πθ(yl|x) - γ)

Since the log probabilities passed to simpo_loss are already length-normalized (average log prob), the loss simplifies to:

logits = β * policy_chosen_logps - β * policy_rejected_logps - γ

where γ = gamma_beta_ratio * beta.

Loss Types

  • sigmoid (default): losses = -log σ(logits) * (1 - label_smoothing) - log σ(-logits) * label_smoothing
  • hinge: losses = relu(1 - logits)

Rewards

  • chosen_rewards = β * policy_chosen_logps
  • rejected_rewards = β * policy_rejected_logps

Default Hyperparameters

  • β = 2.0
  • gamma_beta_ratio = 0.25 (so γ = 0.5)
  • label_smoothing = 0.0
  • loss_type = "sigmoid"
Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 3d ago First seen · 40 lines · 22 tokens per session scan A 744a29e01e5d

Subscribe to this mod's changes

simpo-loss is a skill published in the GitHub repository cxcscmu/SkillLearnBench (83 stars, last pushed 1mo ago), licensed MIT. It adds 22 tokens to every session and 334 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-03.

Related

Other skills, from other repositories

gpt-multimodal

Analyze images and multi-frame sequences using OpenAI GPT series.

benchflow-ai/skillsbench · 18 tokens

agentsop-crewai

SOP for building multi-agent systems with CrewAI — role-based collaboration, sequential/hierarchical processes, Flows, memory, delegation. Use when modeling agent teams with clear roles and task pipelines.

agentsope/SkillAlchemy · 45 tokens

agentsop-dify

SOP for building LLM applications on Dify — visual workflow + chatflow + agent + RAG knowledge base + plugin marketplace + observability, self-hostable. Use when shipping LLM apps fast with a "no-code to pro-code" gradient, especially when non-engineers need to co-author the flow.

agentsope/SkillAlchemy · 69 tokens

agentsop-dspy

Operating SOP for DSPy (Stanford NLP) — the declarative framework for "programming, not prompting" language models. Activate when the user says any of: "use DSPy", "compile a prompt", "optimize prompts/programs", "MIPRO/MIPROv2", "BootstrapFewShot", "GEPA", "Signatures + Modules", "teleprompter", "auto-tune prompts…

agentsope/SkillAlchemy · 157 tokens

agentsop-framework-selection

Neutral, framework-agnostic decision tree for project kickoff: "which agent / RAG / LLM framework should I reach for?" Synthesizes the ecosystem sections of 7 landmark-project SOPs (LangGraph, LlamaIndex, DSPy, CrewAI, vLLM, Aider, Dify) into one layered rubric. Core stance: frameworks are LAYERS, not competitors — a…

agentsope/SkillAlchemy · 189 tokens

agentsop-hybrid-retrieval

Enhancement-overlay SOP for adding sparse (BM25 / keyword) retrieval alongside dense (embedding) retrieval. Activate when a calling agent is building, reviewing, or debugging a retrieval pipeline whose corpus contains exact-match tokens — identifiers, error codes, SKUs, API/function names, proper nouns, citations…

agentsope/SkillAlchemy · 173 tokens