---
format: "aidr-story-markdown/v1"
id: "ad890ab2b3c6d5ae348de89cc5e0f204f086b0a6a3186a4557a48d11280278f1"
canonical_url: "https://aidr.today/ad890ab2?lang=en"
title: "Whitecircle Launches Halo Framework With 2.8x Open Model Throughput"
lang: "en"
requested_lang: "en"
available_langs: ["en","vi"]
translation_fallback: null
fallback_fields: []
published_at: "2026-09-22T05:47:57.000Z"
category: "Infra"
topics: ["whitecircle","halo","throughput","memory","open-source"]
source_urls: ["https://huggingnews.com/ai/whitecircle-launches-halo-framework-with-28x-open-model-throughput-a4b1959b","https://x.com/whitecircle/status/2102087563913609534","https://x.com/yifanzhang_/status/2102182343213543547","https://marketbrief.now/ai/whitecircle-launches-halo-framework-with-28x-open-model-throughput-a4b1959b"]
summary: "The new Halo post-training tool reduces peak memory usage for developers while maintaining native HuggingFace formats. Whitecircle's software delivers a throughput increase of 2.8x over stock TRL and is available for open source models via GitHub. The framework incorporates support for Self-Distilled Policy Gradient (SDPG) to expand the available methodologies for model fine-tuning. This integration allows users to apply SDPG policy gradient optimization directly within the post-training workflow."
---

# Whitecircle Launches Halo Framework With 2\.8x Open Model Throughput

> [Open the canonical story](<https://aidr.today/ad890ab2?lang=en>)

**Published:** 2026-09-22T05:47:57.000Z
**Category:** Infra
**Topics:** whitecircle, halo, throughput, memory, open\-source

## Summary

The new Halo post\-training tool reduces peak memory usage for developers while maintaining native HuggingFace formats\. Whitecircle's software delivers a throughput increase of 2\.8x over stock TRL and is available for open source models via GitHub\. The framework incorporates support for Self\-Distilled Policy Gradient \(SDPG\) to expand the available methodologies for model fine\-tuning\. This integration allows users to apply SDPG policy gradient optimization directly within the post\-training workflow\.

## Sources

- [Story source](<https://huggingnews.com/ai/whitecircle-launches-halo-framework-with-28x-open-model-throughput-a4b1959b>)
- [Story source](<https://x.com/whitecircle/status/2102087563913609534>)
- [Supporting source](<https://x.com/yifanzhang_/status/2102182343213543547>)
- [Story source](<https://marketbrief.now/ai/whitecircle-launches-halo-framework-with-28x-open-model-throughput-a4b1959b>)