来自Polymathic AI，普林斯顿大学，USC ISI,NYU和Flatiron研究所的研究人员拥有发布 FlashMap. FlashMap投机解码器用连续流图来取代离散扩散起草器。 它在1个前传中起草一个完整的16个地块。 那块地上的木桩彼此一致在Qwen3-8B上，研究小组报告GSM8K上最高可达6.14x无损失快取。 Qwen3-8B和Qwen3-4B的起草器重量在Hugging Face上，有vLLM和SGLang插件。
TL; DR 数据
- 大小：0.5B-参数起草器(500M为Qwen3-8B,475M为Qwen3-4B)；草稿块为K=16个令牌。
- 运行在：与目标型号相同的GPU. 纸张快取法使用1NVIDIA H200 NVL，以1为同位素，在bf16中。
- 性能：在Qwen3-8B测试速度最快的起草器，平均比再训练的DFlash高约7%。
- 最佳：在0温下，在GSM8K上加速6.14x(接受长度为7.69令牌).
- 最差：2.28x在气温1时在阿尔帕卡上加速，在Qwen3-4B上(温度为0)，它在9个任务中的4个任务上跟踪了DFlash.
什么是闪光地图？
FlashMap是一个投机性的解码起草器，它将高斯噪声映射到1个符号块。 一个小起草者提出 K 令牌。 目标模型随后检查了1个平行通过中的所有K并保留了最长的匹配前缀。 产出与目标完全相同。
速度取决于两件事。 有多少已起草的代币被接受，而且起草得多么便宜。 EAGLE-3的草稿自动递减，所以成本随着块长而增加。 DFlash在1个传出区块中起草整块。
为什么传播起草者会失去接受？
DFlash 独立预测区块中的每个槽。 研究团队称此为因子化近似。 想象一种“纽约”或“San diego”的续作，其概率相当高。 独立起草人提出“新死神”或“桑·约克”一半时间。 目标拒绝在2号档位的人。
FlashMap基于流图语言模型。 每张噪音样本图到一整块噪音为每个槽共同选择模式。
FlashMap 如何工作？
起草器为6块放送变压器(DiT)，嵌入维度为2048和16个注意力头。 它读取了目标层1、9、17、25和33的隐藏状态。 嵌入矩阵和LM头被复制出目标并保持被冻结。
培训分为两个阶段：
- 流动教师：在一热的符号块上持续流动，受过交叉进取训练。 一个时间再参数化处理Qwen3的151,936-token词汇。
- 流出地图学生：半群分馏使流出地变成2次分相。 它在1个步骤中从噪音跳到数据。
两名起草者都从Open-PerfectBlend接受500K样本的训练。 H200s上，所有型号的计算总时数约为4 600GPU小时。
FlashMap如何比较基准测试？.
研究团队在相同的数据和设置上对EAGLE-3和DFlash进行了再培训。 测试涵盖9个基准测试，跨越了数学，代码和聊天，温度为0和1.
- 速度：1- step FlashMap 是每个Qwen3-8B任务上最快的起草器，在两种温度下。
- 精确度：在18个设置中的15个设置中，1步FlashMap在接受长度上击打DFlash. 它有大约一半的参数。
- 聊天收益最大：在0温度下，比DFlash的得分在数学上为5.5%，聊天为11.1%。
- 2步模式：18个设置中的18个设置的接收长度比DFlash多出10%。
- EAGLE-3：在一号温度下，它的接受长度要高9%，但运行速度要慢35%。
洗牌测试检查草稿是否使用真正的间接结构。 在2步，FlashMap恢复了16步教师的大部分联合结构收益。 在一步，它只保留了一小部分。 起草出4个噪声样本作为一棵树融资，平均速度为11%.
比较：FlashMap 与其他投机解码起草器
平均速度由Marktechpost从FlashMap纸张图1(c)中计算出。 匹配行使用作者的再训练基线。
如何部署闪光地图？
闪光图包只需要火炬和变压器QQ4.56. 用uv pip安装闪光图安装它。 一个 spec generate 调用在 1 GPU 上运行起草器和目标一起运行。
对于服务，项目页面给出了两种途径：
- vLLM 0.31：将起草器通过 -- speculative-config中以 num speculative tokens 设置为 16 个。
- SGLang 0.5.20：使用 -- spective-algorithm DSPARK 与 FlashMap 草稿路径。
设置 num steps=2 以交换某种速度以获得更高的接受。 对于温度为0以上的采样，vLLM需要"draw sample 方法":"概率".
关键外卖
- FlashMap在1个通过中联合起草16个符号并附有连续流图。
- Qwen3-8B上最高可达6.14x无损失快取，使用0.5B起草器。
- 速度快于所有9个Qwen3-8B 基准测试的再训练DFlash.
- 两步模式使整个街区都能平稳接受。
- 重量为Apache-2.0，代码为MIT，有vLLM和SGLang插件。
请检查access-date=中的日期值 (帮助) 论文，项目页，GitHub回放和模型权重。 并加入我们的通讯。 等等，你在电报里吗？现在你也可以加入我们的电报。
需要和我们合作推广您的 GitHub Repo OR Hugging Face Page OR Product Release OR Webinar等吗？联系我们
Asif Razzaq是Marktechpost AI Media Inc.的首席执行官。 作为一个有远见的企业家和工程师，阿西夫致力于利用人工智能的潜力为社会造福。 他最近的努力是人工智能媒体平台的发布，该平台以深入报道机器学习和深层学习新闻而出名，这些新闻在技术上既可靠又为广大观众所容易理解。 该平台每月有超过200万次观看，表明其受观众欢迎。
本文由 AI宝库自动抓取并翻译，版权归原作者所有；如需查看原文请点下方链接。
Researchers from Polymathic AI, Princeton University, USC ISI, NYU and the Flatiron Institute have released FlashMap. FlashMap speculative decoding replaces the discrete diffusion drafter with a continuous flow map. It drafts a full 16-token block in 1 forward pass. Tokens in that block stay consistent with each other. On Qwen3-8B, the research team reports up to 6.14x lossless speedup on GSM8K. Drafter weights for Qwen3-8B and Qwen3-4B are on Hugging Face, with vLLM and SGLang plugins.
TL;DR
- Size: 0.5B-parameter drafter (500M for Qwen3-8B, 475M for Qwen3-4B); drafts blocks of K = 16 tokens.
- Runs on: the same GPU as the target model. Paper speedups use 1 NVIDIA H200 NVL at concurrency 1, in bf16.
- Performance: fastest drafter tested on Qwen3-8B, about 7% above a retrained DFlash on average.
- Best: 6.14x speedup on GSM8K at temperature 0 (acceptance length 7.69 tokens).
- Worst: 2.28x speedup on Alpaca at temperature 1. On Qwen3-4B (temperature 0), it trails DFlash on 4 of 9 tasks.
What is FlashMap?
FlashMap is a speculative decoding drafter that maps Gaussian noise to a token block in 1 step. A small drafter proposes K tokens. The target model then checks all K in 1 parallel pass and keeps the longest matching prefix. Output stays exactly what the target would produce.
Speed depends on 2 things. How many drafted tokens get accepted, and how cheaply they were drafted. EAGLE-3 drafts autoregressively, so cost grows with block length. DFlash drafts the full block in 1 pass with block diffusion.
Why do diffusion drafters lose acceptance?
DFlash predicts each slot in the block independently. The research team calls this the factorized approximation. Picture a continuation that is “new york” or “san diego” with equal odds. An independent drafter proposes “new diego” or “san york” half the time. The target rejects those at slot 2.
FlashMap builds on Flow Map Language Models. Each noise sample maps to 1 full block. The noise picks the mode for every slot jointly.
How does FlashMap work?
The drafter is a 6-block diffusion transformer (DiT), with embedding dimension 2048 and 16 attention heads. It reads hidden states from target layers 1, 9, 17, 25 and 33. Embedding matrix and LM head are copied from the target and stay frozen.
Training runs in 2 stages:
- Flow teacher: a continuous flow over one-hot token blocks, trained with cross-entropy. A time reparameterization handles Qwen3’s 151,936-token vocabulary.
- Flow map student: semigroup distillation turns the flow into a 2-time denoiser. It jumps from noise to data in 1 step.
Both drafters train on 500K samples from Open-PerfectBlend. Total compute across all models and ablations was about 4,600 GPU hours on H200s.
How does FlashMap compare on benchmarks?
The research team retrained EAGLE-3 and DFlash on the same data and settings. Tests cover 9 benchmarks across math, code and chat, at temperatures 0 and 1.
- Speed: 1-step FlashMap is the fastest drafter on every Qwen3-8B task, at both temperatures.
- Accuracy: 1-step FlashMap beats DFlash on acceptance length in 15 of 18 settings. It has about half the parameters.
- Chat gains are largest: at temperature 0, gains over DFlash are 5.5% on math, 11.1% on chat.
- 2-step mode: acceptance length beats DFlash in 18 of 18 settings, by 10% on average.
- EAGLE-3: at temperature 1, it has 9% higher acceptance length but runs 35% slower.
A shuffle test checks whether drafts use real inter-token structure. At 2 steps, FlashMap recovers most of the joint-structure gain of a 16-step teacher. At 1 step, it keeps only a small fraction. Drafting 4 noise samples as a tree raised speedup by 11% on average.
Comparison: FlashMap vs other speculative decoding drafters
Average speedups are computed by Marktechpost from Figure 1(c) of the FlashMap paper. Matched rows use the authors’ retrained baselines.
How do you deploy FlashMap?
The flashmap package needs only torch and transformers>=4.56. Install it with uv pip install flashmap. A spec_generate call runs drafter and target together on 1 GPU.
For serving, the project page gives 2 paths:
- vLLM 0.31: pass the drafter in --speculative-config with num_speculative_tokens set to 16.
- SGLang 0.5.20: use --speculative-algorithm DSPARK with the FlashMap draft path.
Set num_steps=2 to trade some speed for higher acceptance. For sampling above temperature 0, vLLM needs "draft_sample_method": "probabilistic".
Key Takeaways
- FlashMap drafts 16 tokens jointly in 1 pass with a continuous flow map.
- Up to 6.14x lossless speedup on Qwen3-8B, using a 0.5B drafter.
- Faster than a retrained DFlash on all 9 Qwen3-8B benchmarks.
- 2-step mode keeps acceptance flat across the block.
- Weights are Apache-2.0, code is MIT, with vLLM and SGLang plugins.
Check out the paper, project page, GitHub repo and model weights. Also, feel free to follow us on Twitter and don’t forget to join our 150k+ML SubReddit and Subscribe to our Newsletter. Wait! are you on telegram? now you can join us on telegram as well.
Need to partner with us for promoting your GitHub Repo OR Hugging Face Page OR Product Release OR Webinar etc.? Connect with us
Asif Razzaq is the CEO of Marktechpost AI Media Inc.. As a visionary entrepreneur and engineer, Asif is committed to harnessing the potential of Artificial Intelligence for social good. His most recent endeavor is the launch of an Artificial Intelligence Media Platform, Marktechpost, which stands out for its in-depth coverage of machine learning and deep learning news that is both technically sound and easily understandable by a wide audience. The platform boasts of over 2 million monthly views, illustrating its popularity among audiences.
Auto-fetched and translated by AI Gem Box. All rights belong to the original author.