@LiorOnAI: An open-source fix for one of the most common reasoning model failure modes. One of the biggest AI trends this year isn…
Summary
Liquid AI releases Antidoom, an open-source method that fine-tunes reasoning models to break repetitive token loops (doom loops), reducing failure rates from ~23% to 1% on Qwen3.5-4B without retraining or RL.
View Cached Full Text
Cached at: 07/07/26, 11:38 PM
An open-source fix for one of the most common reasoning model failure modes.
One of the biggest AI trends this year isn’t larger models. It’s systematically removing failure modes.
Reasoning models sometimes get stuck repeating the same token sequence (“Wait…”, “So…”, “Alternatively…”) until they exhaust the context window.
Antidoom finds the single token that starts the loop and fine-tunes the model to prefer alternative next tokens at that position. It doesn’t retrain the model from scratch, use RL, or teach the model new knowledge.
On Qwen3.5-4B, doom loops dropped from 22.9% to 1%, and benchmark scores increased because the model stopped getting trapped before producing answers it was already capable of generating.
Liquid AI (@liquidai): Today we release Antidoom, an open-source method that removes a common failure mode in reasoning models: the doom loop.
Doom-loop rates before and after, with eval scores up across the board:
> Early LFM2.5-2.6B checkpoint: 10.2% → 1.4%
> Qwen3.5-4B: 22.9% → 1% (greedy
Similar Articles
@helloiamleonie: Working with the @liquidai team on these engineering blogs is just so much fun! Here's what we've been working on: Reas…
The post explains doom loops in reasoning models where the model repeats tokens like 'Wait' until the context fills up, and introduces FTPO (Final Token Preference Optimization) as a training-time fix. The associated Antidoom tool reduces doom loop rates significantly (e.g., from 22.9% to 1% on Qwen3.5-4B).
@maximelabonne: New training technique to reduce doom loops! We applied it to LFM2.5-2.6B (SFT checkpoint) Qwen3.5-4B. By reducing doom…
Liquid AI releases Antidoom, an open-source method to reduce doom loops in reasoning models, applied to LFM2.5-2.6B and Qwen3.5-4B, significantly lowering doom-loop rates and improving eval scores.
Reducing Doom Loops with Final Token Preference Optimization
Liquid AI introduces Antidoom, a method using Final Token Preference Optimization to reduce repetitive doom loops in small reasoning models during inference, cutting loop rates from 10.2% to 1.4% while improving evaluation scores.
@LiorOnAI: The bottleneck is no longer just compute or parameters. It’s access to high-quality reasoning traces. Once millions of …
Lior On AI argues that high-quality reasoning traces are the new bottleneck in AI. A team distilled 2.3M reasoning traces from Claude Fable 5 into Qwen3-4B, achieving perfect self-consistency and zero hallucination variance, and open-sourced the result.
Reasoning Can Be Restored by Correcting a Few Decision Tokens
This paper shows that the reasoning gap between base LLMs and large reasoning models is concentrated on a small set of early planning tokens. It introduces disagreement-guided token intervention, where replacing only those critical tokens with a reasoning model's outputs allows a base model to nearly match the reasoning model's performance.