Fetching from the wire…
01
02
03
04
05
06
07
08
Public story · 2026-07-25 · high
Rust draws 2 to 3 times more dedicated circuitry than Python, and Qwen keeps its concept circuits at layers 17-19 while DeepSeek keeps them at 6-7.
Why now: This circuit comparison is part of the July 25, 2026 research briefing.
Two code models agree on which grammar concepts deserve dedicated circuitry but disagree completely on where those circuits sit, per Piotr Wilam's comparison of Qwen2.5-Coder-7B and DeepSeek-Coder-V1-6.7B.
That's a problem for anyone comparing code models by layer number. A probe tuned to one model's layer won't find the matching circuit in another. Qwen clusters this work at layers 17-19; DeepSeek does the same job at layers 6-7.
The agreement on concepts is strong. Spearman correlation between the two models hits 0.638 for Python and 0.673 for Rust, both p < 1e-7. The comparison covers 58 Python concepts and 57 Rust concepts, tagged the same way across all four model-language pairs.
Rust also draws more machinery than Python in both models, 2 to 3 times more concept-specific circuitry per construct. Qwen goes further, binding nine Rust type-and-trait keywords into one neuron cluster, a Jaccard score of 0.535 against a null baseline of 0.112.
The paper doesn't say why DeepSeek compresses the work into 2 layers while Qwen spreads it across 3. It also doesn't say whether the split holds outside these two model families. Its own conclusion is blunt: interpretability tooling that assumes layer positions transfer between models is measuring the wrong thing.
Each link below shares sources, entities, or timing with this story.
Qwen benchmarked against Claude / Shared entities / Same source / Shared topic / Earlier coverage
Linked by a graph relationship (Qwen benchmarked against Claude); both cover Code, Piotr Wilam; cite the same source (Piotr Wilam).
Anthropic criticizes Qwen / Shared entities / Earlier coverage / Tension
Linked by a graph relationship (Anthropic criticizes Qwen); both cover Python, Rust, Which; earlier Python coverage from 2026-07-20.
Anthropic criticizes Qwen / Shared entities / Earlier coverage
Linked by a graph relationship (Anthropic criticizes Qwen); both cover DeepSeek, Qwen, Rust; earlier DeepSeek coverage from 2026-05-01.
Qwen benchmarked against Claude / Shared entities / Shared topic / Earlier coverage
Linked by a graph relationship (Qwen benchmarked against Claude); both cover Qwen, Which; overlapping topics (code, model).
Anthropic criticizes Qwen / Shared entities / Shared topic / Earlier coverage
Linked by a graph relationship (Anthropic criticizes Qwen); both cover Python, Rust; overlapping topics (code, model).
Linked by a graph relationship (Anthropic criticizes Qwen); both cover DeepSeek, Qwen; overlapping topics (layer, model).
Ollama supports Qwen / Shared entities / Shared topic / Earlier coverage
Linked by a graph relationship (Ollama supports Qwen); both cover Python, Rust; overlapping topics (python, rust).
Qwen benchmarked against Claude / Shared entity: Which / Shared topic / Earlier coverage / Tension
Linked by a graph relationship (Qwen benchmarked against Claude); both cover Which; overlapping topics (code, model).