百度“无限OCR”通过将记忆模拟为人类遗忘，单次处理数十页文档

Baidu researchers have built an OCR model that handles dozens of document pages in a single inference pass, keeping memory use and speed constant regardless of text length. A redesigned attention mechanism makes this possible. No current OCR model handles more than about ten pages in a single pass, the Baidu researchers write in their technical report. The bottleneck is the KV cache, a buffer where a language model stores all previously processed tokens during generation so it can look them up later. Current end-to-end systems use a language model as their decoder, so this buffer grows with every new line of text. That drives up memory use and steadily slows generation. In practice, systems get around the problem with a loop that processes each document page by page, resetting the cache after every step. Baidu frames the problem with a human analogy. Someone copying a book doesn't re-read everything they've already written. They keep their eyes on the source, the last few characters they wrote, and the next one to put down. Older passages fade through a kind of soft forgetting. The researchers want Unlimited OCR to mimic that pattern. It works through what the team calls Reference Sliding Window Attention (R-SWA). Each generated token still sees all reference tokens, the visual image tokens and the prompt. But when it comes to previously generated output, it only looks back at the last 128 tokens. That keeps the KV cache constant throughout the entire process instead of growing linearly with output length. Standard sliding window attention would also subject visual tokens to ongoing state changes, gradually blurring image features and degrading recognition. R-SWA exempts visual tokens from these transitions. They're encoded once and stay unchanged. The KV cache works as a queue where each new token pushes out the oldest one. With standard multi-head attention, memory use grows without bound as token count rises. R-SWA caps it at the fixed sum of prefix length and window size. Unlimited OCR builds on the open-source Deepseek OCR model . Baidu keeps its DeepEncoder and pairs it with a mixture-of-experts architecture with three billion parameters, of which only about 500 million are active during inference. The DeepEncoder compresses a 1024-by-1024-pixel PDF image down to 256 tokens. Two resolution modes carry over. "Base" mode handles multi-page documents, and "Gundam" mode uses dynamic resolution for single pages. Every standard attention layer in the decoder was swapped out for R-SWA. Training used about two million document samples, split 9-to-1 between single-page and multi-page data. Paddle OCR handled annotation for single pages. Multi-page data was built synthetically by stitching single pages together into documents ranging from two to 50 pages. All data was packed into sequences of 32,000 tokens; training ran for 4,000 steps on 8 times 16 Nvidia A800 GPUs. The DeepEncoder stayed frozen, and only the language model parameters were updated. Unlimited OCR scores 93 percent overall on the OmniDocBench v1.5 document benchmark, six percentage points above the Deepseek OCR baseline, according to the authors. The benchmark measures several sub-tasks. Pure text recognition error rate, measured as edit distance (the number of corrections needed per character), drops slightly. Table structure recognition improves more sharply, by nearly six percentage points. On the newer v1.6 version, the model hits 93.92 percent, putting it at the top of the end-to-end system rankings.