包裹。 接下来:“We study 1,260,967 blind votes cast from May 2023 to August 2024 on pairs of models whose versions are fixed: dated vendor snapshots and models with fixed open weights. For the same pair of unchanged models, the share of votes saying "both are bad" rises by 1.71 percentage points a month on a mean of 17.1%, and by 1.91 points a month in 2024. In fourteen pairs observed often in both 2023 and 2024, it rose in every one, from 15.5% to 23.3%.”这部分是研究数据和初步结果,也是
?或者有没有小节标题?原文没有小标题,所以可能需要用
吗?看规则2,长文用
/
划分小节,原文是一段,但内容有不同部分,可能需要划分。比如第一部分是背景和方法,第二部分是研究结果,第三部分是趋势和普遍性,第四部分是结论。 等一下,原文的结构: 1. 背景:用户抱怨,两种可能,研究平台介绍。 2. 研究数据:1,260,967 votes,时间段,模型类型(固定版本)。 3. 结果:同一对模型,“both are bad”的比例每月上升,2024年更高;14对模型都上升。 4. 排名趋势:2024年,强模型的决胜投票比例不变。 5. 漂移的普遍性:同一选民、同一对模型、不同到达群体、提示控制、多语言、不同模型类型、界面变化都成立;跨模型共享漂移。 6. 结论:平台层面,平均模型对2023年的得分预期,但“both are bad”的比例回到起点;模式符合标准跟随类别进步,静止的产品被判断为下降;绝对评价漂移,相对比较稳定,对排行榜和训练用人类反馈重要。 所以需要划分小节吗?比如:
研究背景与平台介绍
Users often complain that AI assistants such as ChatGPT are getting worse. Some versions do change, but human evaluation leaves a second possibility untested: the people judging may change. We separate the two on Chatbot Arena, a public platform where people compare answers from two anonymous large language models (LLMs) and pick the better one or say both are bad.
研究数据与初步结果
We study 1,260,967 blind votes cast from May 2023 to August 2024 on pairs of models whose versions are fixed: dated vendor snapshots and models with fixed open weights. For the same pair of unchanged models, the share of votes saying "both are bad" rises by 1.71 percentage points a month on a mean of 17.1%, and by 1.91 points a month in 2024. In fourteen pairs observed often in both 2023 and 2024, it rose in every one, from 15.5% to 23.3%.
排名趋势与漂移的普遍性
The ranking does not trend: in 2024, the stronger model's share of decisive votes is flat. The rise holds within the same voter and on the same pair, and across arrival cohorts from their first recorded vote. It holds with prompt controls, in every major language, for dated vendor snapshots, open-weight models and models with stable answer length, and between interface changes. The drift is shared across models: month effects estimated from other models' pairs remove it from each model's own pairs.
研究结论
Platform-wide, the average model would be expected to score 78% against the 2023 average, yet the share of "both are bad" votes ended near where it began. The pattern fits a standard that follows the category's progress, so a product that stands still is judged as declining. Absolute verdicts on AI drifted while relative comparisons held steady, an evaluation drift that matters for LLM leaderboards and for the human feedback used to train models.
这样对吗?原文没有小标题,但根据内容划分小节是允许的,规则2说长文用/
划分小节。那这样是符合的。然后检查有没有需要用的句子,比如重要的句子,比如“the share of votes saying "both are bad" rises by 1.71 percentage points a month on a mean of 17.1%, and by 1.91 points a month in 2024.”或者“Absolute verdicts on AI drifted while relative comparisons held steady”这些可以加吗?规则3说重要句子可用,那比如“the share of votes saying "both are bad" rises by 1.71 percentage points a month on a mean of 17.1%, and by 1.91 points a month in 2024.”可以加吗?或者“Absolute verdicts on AI drifted while relative comparisons held steady”是结论里的重要句子,可以加。 比如:
研究数据与初步结果
We study 1,260,967 blind votes cast from May 2023 to August 2024 on pairs of models whose versions are fixed: dated vendor snapshots and models with fixed open weights. For the same pair of unchanged models, the share of votes saying "both are bad" rises by 1.71 percentage points a month on a mean of 17.1%, and by 1.91 points a month in 2024. In fourteen pairs observed often in both 2023 and 2024, it rose in every one, from 15.5% to 23.3%.
研究结论
Platform-wide, the average model would be expected to score 78% against the 2023 average, yet the share of "both are bad" votes ended near where it began. The pattern fits a standard that follows the category's progress, so a product that stands still is judged as declining. Absolute verdicts on AI drifted while relative comparisons held steady, an evaluation drift that matters for LLM leaderboards and for the human feedback used to train models.
这样对吗?然后检查有没有遗漏信息,比如原文的“dated vendor snapshots”“fixed open weights”“arrival cohorts”“prompt controls”“interface changes”“month effects estimated from other models' pairs”这些都保留了。有没有列表?原文没有列表,所以不用
。有没有引用?没有。有没有代码?没有。所以结构应该是:
研究背景与平台介绍
Users often complain that AI assistants such as ChatGPT are getting worse. Some versions do change, but human evaluation leaves a second possibility untested: the people judging may change. We separate the two on Chatbot Arena, a public platform where people compare answers from two anonymous large language models (LLMs) and pick the better one or say both are bad.
研究数据与初步结果
We study 1,260,967 blind votes cast from May 2023 to August 2024 on pairs of models whose versions are fixed: dated vendor snapshots and models with fixed open weights. For the same pair of unchanged models, the share of votes saying "both are bad" rises by 1.71 percentage points a month on a mean of 17.1%, and by 1.91 points a month in 2
研究背景与平台介绍
Users often complain that AI assistants such as ChatGPT are getting worse. Some versions do change, but human evaluation leaves a second possibility untested: the people judging may change. We separate the two on Chatbot Arena, a public platform where people compare answers from two anonymous large language models (LLMs) and pick the better one or say both are bad.
研究数据与初步结果
We study 1,260,967 blind votes cast from May 2023 to August 2024 on pairs of models whose versions are fixed: dated vendor snapshots and models with fixed open weights. For the same pair of unchanged models, the share of votes saying "both are bad" rises by 1.71 percentage points a month on a mean of 17.1%, and by 1.91 points a month in 2