AI & ML
Computed Permutation Trellis Coding: Near-Optimal Quantization from Two Multiplications and a 64-Byte Table
arXiv math.IT0 views
arXiv:2611.22019v1 Announce Type: new Abstract: We propose Computed Permutation Trellis Coding (CPTC), a low-bit quantization format for tensors written once and read many times. Formats that round each value on its own, such as INT4 and FP4, waste a large part of their bits against the information-theoretic ideal. Trellis codes recover this loss by encoding values jointly, but prior decoders either need a large lookup table for their best quality or compute values with a hash that restricts what they can decode. CPTC's decoder multiplies its state by two coefficients and adds the top bits of the products to index a 64-byte table. Because that sum is not uniformly distributed, the rarely used ends of a small table can hold values far out in the tails. One decoder serves every bit width from 1 to 6, reproduces the standard integer and floating-point formats exactly, and comes within 0.07 bits of the ideal at 4 bits. It matches a random-permutation trellis code whose table is thousands of times larger, and exceeds every QTIP codebook. In LLM weight quantization without fine-tuning, CPTC has the lowest KL divergence to the unquantized model at every tested bit width on Llama 3.1 8B and 70B Instruct against QTIP's codebooks (by up to 27%). The decoder has been built into the AWS Trainium 4 AI accelerator, where it runs at the throughput of the native FP4 and FP8 formats for a negligible fraction of the systolic array's area. Fused encoding kernels quantize an 8B model's linear layers in under two minutes on one GPU and, in cycle-level simulation, on one Trainium 4 chip.
Source:https://arxiv.org/abs/2611.22019PreviousAn upper bound on non-overlapping codesNextSubsystem Stabilizer $(r,\delta)$-Locally Recoverable Codes: Characterization, Bounds and Constructions
Related posts

Sources: Nvidia made two separate introductions in 2022 connecting Alibaba with alleged Chinese spy Alice Huang, actively helping Megaspeed land Alibaba (The Information)
0Techmeme
Show HN: I cut an LLM's screen-watching cost to ~138 tokens per step
0Hacker News New
Codeberg's Domain Was Suspended
0Hacker News New
Gemini AI agents repeatedly bypassed safeguards to attack real website
0Hacker News New
A new AI agent builder. No-code visual builder with customer support platform
0Hacker News New
Jacob Tsimerman on AI can supercharge mathematics [video]
0Hacker News New