simpo-loss-function SKILL.md skillA
simpo-loss-function from cxcscmu/skilllearnbench is an agent skill: a SKILL.md folder an AI agent loads when a task calls for it.
mdr add cxcscmu/skilllearnbench/simpo-loss-function@git:20260424.d50cf5aPins git:20260424.d50cf5a in mdr.lock. Pin the fingerprint instead.
- Audit grade
- A, 17 of 17 checks passed
- Kind
- SKILL.md skill
- Repository
- cxcscmu/skilllearnbench
- Path
skills/b1-one-shot-claude-haiku-4-5/nlp-paper-reproduction/simpo-loss-function/SKILL.md- Versions recorded
- 1
- Latest version
- git:20260424.d50cf5a, committed Apr 24, 2026
- Size
- 2,739 bytes
- Repository stars
- 91
- License
- MIT
Implement SimPO loss with length-normalized rewards and target margin.
Collected from public GitHub. Every version is saved exactly as it was and never changes. Install it at an exact version with the mdr tool, and every file is checked before it reaches your agent. The main file is checked against the SHA-256 fingerprint recorded here, and the others against the fingerprints GitHub keeps for its source commit. The automatic checks below grade the latest version, and the same checks always give the same file the same grade.
What the file says
# SimPO Loss Function Implementation ## Overview SimPO (Simple Preference Optimization) implements a preference optimization objective that uses length-normalized average log probability as an implicit reward, with a target reward margin component. ## Key Formula ``` L_SimPO(πθ) = -E_(x,yw,yl)~D log σ(β/|yw| log πθ(yw|x) - β/|yl| log πθ(yl|x) - γ) ``` ## Components ### 1. Length-Normalized Reward - **Formula**: `r_SimPO(x, y) = β/|y| * log πθ(y|x)` - **Purpose**: Average log probability per token, prevents length bias - **Why**: Aligns training with generation metric (which uses average log likelihood for beam search) ### 2. Bradley-Terry Objective - **Formula**: `p(yw ≻ yl | x) = σ(r(x, yw) - r(x, yl) - γ)` - **Purpose**: Probabilistic ranking between winning and losing responses - **Function**: σ is sigmoid function ### 3. Target Reward Margin (γ) - **Purpose**: Ensure reward difference exceeds a target threshold - **Effect**: Improves generalization by enforcing margin between classes - **Typical range**: 0.5 to 1.5 ## Implementation Details ### Computing Log Probabilities ```python # log_probs shape: (batch_size, seq_len) …
Read the whole file at its exact version.
How to install
mdr add cxcscmu/skilllearnbench/simpo-loss-function@sha256:dd61897e5dbb4f07The line at the top of this page pins the label git:20260424.d50cf5a, which is what most people want. This one pins the sha256 fingerprint of the contents instead, so it survives a label being moved. Either way the exact version is written to mdr.lock, and mdr install downloads and checks those contents again, so it installs them exactly or stops with an error.
[](https://markdownregistry.com/a/art_4tkoaj2ev4c7n4dd)
3 badge image fetches in 30 days. What the badge shows.
Versions
The registry has recorded 1 version of this file, committed on Apr 24, 2026.
Audit of the latest version
- pass: Frontmatter block present
- pass: Frontmatter declares a name
- pass: Frontmatter declares a description
- pass: Size between 200 bytes and 200 KB (2739 bytes)
- pass: No zero-width or bidi control characters
- pass: No instruction hidden inside an HTML comment
- pass: No link to an exfiltration or paste host
- pass: No credential-shaped string
- pass: No instruction to send local credentials anywhere
- pass: No text hidden with inline styles
- pass: No prompt-injection phrasing
- pass: No curl or wget piped into a shell
- pass: No recursive delete of root, home or parent
- pass: No instruction to read or print local credentials
- pass: No base64 blob over 200 characters
- pass: No link to a raw IP address
- pass: No script tag
Source
cxcscmu/skilllearnbench · 91 stars · license MIT · pushed 2026-10-02 · branch main
API
GET https://markdownregistry.com/api/v1/artifacts/art_4tkoaj2ev4c7n4dd GET https://markdownregistry.com/api/v1/resolve?ref=cxcscmu/skilllearnbench/simpo-loss-function GET https://markdownregistry.com/api/v1/blob/dd61897e5dbb4f0741af9e3cfa22eb1d90b9e442ef4443addec022896026726c
Your AI assistant does the legwork. You hear about the deals worth your word. Hand yours the standing instructions at modelranch.com and it joins the network that reads files like this one.
Guides: the SKILL.md format and its required fields · what to check before you install a skill · how to pin a skill to an exact version · where to find agent skills · keeping a team on the same skills · which agent file does what.