hyperloom-qwen3-14b-fp8-12h-forge SKILL.md skillB
hyperloom-qwen3-14b-fp8-12h-forge from amd-agi/hyperloom is an agent skill: a SKILL.md folder an AI agent loads when a task calls for it.
mdr add amd-agi/hyperloom/hyperloom-qwen3-14b-fp8-12h-forge@git:20261106.2f99e29Pins git:20261106.2f99e29 in mdr.lock. Pin the fingerprint instead.
- Audit grade
- B, 16 of 17 checks passed
- Kind
- SKILL.md skill
- Repository
- amd-agi/hyperloom
- Path
examples/hyperloom-qwen3-14b-fp8-12h-forge/SKILL.md- Versions recorded
- 19
- Latest version
- git:20261106.2f99e29, committed Nov 6, 2026
- First version recorded
- Sep 9, 2026
- Size
- 15,074 bytes
- Repository stars
- 262
- License
- a license file GitHub does not identify
Run a 12-hour Hyperloom Qwen3-14B-FP8 optimization session on SGLang, vLLM or ATOM with the per-kernel KernelForge backend instead of GEAK. Use when the user wants the medium-length Hyperloom demo and has asked for the forge kernel backend.
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
# Hyperloom Qwen3-14B-FP8 12h Run (Forge Kernel Backend)
Load `.env` with the preamble below and resolve `HYPERLOOM_SKILL_PATH`. Read and follow the optimizer skill at `@${HYPERLOOM_SKILL_PATH}` before launching. If `HYPERLOOM_SKILL_PATH` is missing, fall back to `@hyperloom/inference_optimizer/SKILL.md` (wheel install) or `@src/hyperloom/inference_optimizer/SKILL.md` (source checkout). This skill provides the concrete workload and launch constraints for a 12-hour Qwen3-14B-FP8 demo.
This is the [`hyperloom-qwen3-14b-fp8-12h`](../hyperloom-qwen3-14b-fp8-12h/SKILL.md)
demo with **one** difference: the KERNEL_AGENT phase runs the per-kernel
KernelForge backend instead of GEAK. The workload, budget, and phase split are
identical on purpose, so the two runs stay directly comparable. SGLang and vLLM
follow the sections below; for ATOM (`FRAMEWORK=atom`) follow [ATOM](#atom).
## Execution Shell
Run this in the current Hyperloom workspace before setup or runtime installation.
Repeat it in each new execution shell, including inside Docker; the shared loader
fills dotenv gaps without replacing existing non-empty exports.
```bash
set -e
export REPO_ROOT="$(pwd -P)"
…Read the whole file at its exact version.
How to install
mdr add amd-agi/hyperloom/hyperloom-qwen3-14b-fp8-12h-forge@sha256:7843b3b21c6e9ecfThe line at the top of this page pins the label git:20261106.2f99e29, 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_4jpod2b2acd2p2ac)
4 badge image fetches in 30 days. What the badge shows.
Versions
The registry has recorded 19 versions of this file, between Sep 9, 2026 and Nov 6, 2026, about one every 3 days.
| version | committed | commit | size | audit | |
|---|---|---|---|---|---|
| git:20261106.2f99e29 latest | 2026-11-06 | 2f99e29 | 15,074 B | B | view · diff |
| git:20261103.a5836c9 | 2026-11-03 | a5836c9 | 15,103 B | B | view · diff |
| git:20261026.32e80a9 | 2026-10-26 | 32e80a9 | 15,103 B | B | view · diff |
| git:20261021.fbcd0c3 | 2026-10-21 | fbcd0c3 | 15,103 B | B | view · diff |
| git:20261012.b22d13b | 2026-10-12 | b22d13b | 15,093 B | B | view · diff |
| git:20261009.82e4b51 | 2026-10-09 | 82e4b51 | 14,941 B | B | view · diff |
| git:20261009.976d7a0 | 2026-10-09 | 976d7a0 | 14,941 B | B | view · diff |
| git:20260930.07a30dc | 2026-09-30 | 07a30dc | 15,029 B | B | view · diff |
| git:20260924.28e34b0 | 2026-09-24 | 28e34b0 | 14,003 B | B | view · diff |
| git:20260923.db3cc15 | 2026-09-23 | db3cc15 | 14,057 B | B | view · diff |
| git:20260923.44d8731 | 2026-09-23 | 44d8731 | 14,288 B | B | view · diff |
| git:20260921.b876129 | 2026-09-21 | b876129 | 14,251 B | B | view · diff |
| git:20260917.2079877 | 2026-09-17 | 2079877 | 14,253 B | B | view · diff |
| git:20260917.0d8dd3b | 2026-09-17 | 0d8dd3b | 13,780 B | B | view · diff |
| git:20260917.3fcc929 | 2026-09-17 | 3fcc929 | 13,891 B | B | view · diff |
| git:20260915.97eadf5 | 2026-09-15 | 97eadf5 | 13,499 B | B | view · diff |
| git:20260914.f052bba | 2026-09-14 | f052bba | 13,499 B | B | view · diff |
| git:20260910.60d21b9 | 2026-09-10 | 60d21b9 | 13,499 B | B | view · diff |
| git:20260909.0a8edb1 | 2026-09-09 | 0a8edb1 | 13,454 B | B | view |
Audit of the latest version
- fail: No instruction to read or print local credentials (matched: Read `HYPERLOOM_DOCKER_TARGET_HOST` from `.env)
- pass: Frontmatter block present
- pass: Frontmatter declares a name
- pass: Frontmatter declares a description
- pass: Size between 200 bytes and 200 KB (15074 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 base64 blob over 200 characters
- pass: No link to a raw IP address
- pass: No script tag
Starting a new one? Use a template for SKILL.md on mdpantry, which passes every one of these checks and says what to change in it.
Source
amd-agi/hyperloom · 262 stars · license a license file GitHub does not identify · pushed 2026-11-06 · branch main
API
GET https://markdownregistry.com/api/v1/artifacts/art_4jpod2b2acd2p2ac GET https://markdownregistry.com/api/v1/resolve?ref=amd-agi/hyperloom/hyperloom-qwen3-14b-fp8-12h-forge GET https://markdownregistry.com/api/v1/blob/7843b3b21c6e9ecff81950f48b74f20ed2ca7792dbf8b6139414b6e50d545b52
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.