adamjb111/neuronmono-qwen35-2b-fineweb10m-20261017

neuronmono: Qwen3.5-2B MLP neuron atlas on 10M FineWeb tokens (17 Oct 2026) A neuron = one input to mlp.down_proj (silu(gate·x)·(up·x)); 24 layers x 6,144 = 147,456 neurons, each with a + and a − pole. Text: 24,400 FineWeb (sample-10BT, shard 000_00000, ODC-BY) passages of 512 Qwen3.5 tokens, from document starts. Files tok_A.npy (20,000 x 512, stats set), tok_B.npy (4,000, held-out band samples), tok_C.npy (400, eval) — int32 token ids. stats.pt (torch): top_val/top_idx… See the full description on the dataset page: https://huggingface.co/datasets/adamjb111/neuronmono-qwen35-2b-fineweb10m-202

<iframe src="https://silytics.si/embed/item/ee69a299-84a6-489b-988b-6f046e0d800a" width="420" height="190" style="border:0" loading="lazy" title="Silytics entry"></iframe>

Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 18 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.