Saved in:
| Main Author: | |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2605.21333 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866914583311024128 |
|---|---|
| author | Liu, Ting |
| author_facet | Liu, Ting |
| contents | Natively trained spiking language models struggle to combine Transformer-like language quality, stable multi-domain pre-training, and high activation sparsity. We present SymbolicLight V1, a spike-gated dual-path language model that combines binary Leaky Integrate-and-Fire spike dynamics with a continuous residual stream. Its Dual-Path SparseTCAM module replaces dense self-attention with an exponential-decay aggregation path for long-range memory and a spike-gated local attention path for short-range precision, complemented by a dynamic context-conditioned decoding head and a bilingual tokenizer.
A 194M-parameter SymbolicLight V1 model trained from scratch on a 3B-token Chinese-English corpus reaches held-out validation PPL 8.88-8.93 across four independent runs at >89% per-element activation sparsity. It trails GPT-2 201M by 7.7% in PPL while surpassing GPT-2 124M under the reported comparison. Component ablations at matched 0.5B-token training budgets show that the spike-gated local attention path is the largest contributor, and that replacing LIF dynamics with a deterministic top-k mask at matched sparsity causes a larger degradation, indicating that temporal integration rather than sparsity alone drives performance. We also report a 0.8B-parameter scale-up run trained on 48.8B tokens as evidence of optimization and sparsity preservation, not as a primary quality comparison. Current dense-hardware inference is slower than GPT-2, so neuromorphic deployment is presented as a future sparsity-driven opportunity rather than an achieved hardware speedup. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_21333 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | SymbolicLight V1: Spike-Gated Dual-Path Language Modeling with High Activation Sparsity and Sub-Billion-Scale Pre-Training Evidence Liu, Ting Computation and Language Artificial Intelligence Natively trained spiking language models struggle to combine Transformer-like language quality, stable multi-domain pre-training, and high activation sparsity. We present SymbolicLight V1, a spike-gated dual-path language model that combines binary Leaky Integrate-and-Fire spike dynamics with a continuous residual stream. Its Dual-Path SparseTCAM module replaces dense self-attention with an exponential-decay aggregation path for long-range memory and a spike-gated local attention path for short-range precision, complemented by a dynamic context-conditioned decoding head and a bilingual tokenizer. A 194M-parameter SymbolicLight V1 model trained from scratch on a 3B-token Chinese-English corpus reaches held-out validation PPL 8.88-8.93 across four independent runs at >89% per-element activation sparsity. It trails GPT-2 201M by 7.7% in PPL while surpassing GPT-2 124M under the reported comparison. Component ablations at matched 0.5B-token training budgets show that the spike-gated local attention path is the largest contributor, and that replacing LIF dynamics with a deterministic top-k mask at matched sparsity causes a larger degradation, indicating that temporal integration rather than sparsity alone drives performance. We also report a 0.8B-parameter scale-up run trained on 48.8B tokens as evidence of optimization and sparsity preservation, not as a primary quality comparison. Current dense-hardware inference is slower than GPT-2, so neuromorphic deployment is presented as a future sparsity-driven opportunity rather than an achieved hardware speedup. |
| title | SymbolicLight V1: Spike-Gated Dual-Path Language Modeling with High Activation Sparsity and Sub-Billion-Scale Pre-Training Evidence |
| topic | Computation and Language Artificial Intelligence |
| url | https://arxiv.org/abs/2605.21333 |