sparse-memory-lm
A 21M-parameter LM with a 16.8M-row product-key memory: as good as a 114M dense model and runs with the table on an SSD. Triton kernels for ROCm and CUDA.
Get picks like this daily. The day's top launches, AI/tech news, and a weekly opportunity spotlight — straight to your inbox.
This is 1 of 239 launches in model optimization and inference runtimes — see how it stacks up on momentum and crowding →
727 other launches read as similar to this one →
Details
- External ID
- 1403330296
- Source
- GITHUB
- Company
- —
- Product
- sparse-memory-lm
- Website domain
- github.com
- Launched
- Oct. 3, 2026
- Cohort
- —
- Upvotes
- 10
- Upvotes percentile
- 0.09
- Tags
- amd, language-model, memory-layers, product-key-memory, pytorch, rocm, triton
Enrichment
- Niche
- model optimization and inference runtimes
- Vertical
- Horizontal
- Function
- Model & infra
- Audience
- Developer
- AI stance
- AI-native
- Project type
- Hobby / open-source project
- Normalized one-liner
- memory-augmented language model for resource-constrained hardware
- Manually corrected
- False
Could you build this?
No Developing a custom sparse-memory language model with a 16.8M-row product-key memory and custom Triton GPU kernels running from SSD requires advanced deep learning systems research.
What it would actually take: Requires research-level expertise in transformer architectures, Product-Key Memory (PKM), asynchronous I/O (io_uring/SSD paging), and writing high-performance low-level Triton/CUDA/ROCm kernels to handle sparse memory lookups with minimal latency.
Competitors
Other products that read as similar to this one — 727 launches clear the similarity bar, closest 8 shown.
Attention rank: #653 of 728 (itself plus its competitors, highest first — normalized so YC and Product Hunt are compared fairly).
Launched 335 days after the earliest competitor.
- Long term Memory and 50M token window for LLM · hn · 2026-10-09 · 6 upvotes · similarity 0.53
- llm-compute-allocation-modeling · github · 2026-09-24 · 11 upvotes · similarity 0.53
- dlm-mmd · github · 2026-10-06 · 12 upvotes · similarity 0.52
- Tiny-vLLM · hn · 2026-05-29 · 205 upvotes · similarity 0.52
- Open-source AMDGCN kernels for optimizing LLM inference · hn · 2026-08-25 · 5 upvotes · similarity 0.51
- Minimal LLM Post-Training Experiments on an 8GB GPU (SFT, DPO, GRPO) · hn · 2026-08-01 · 21 upvotes · similarity 0.50
- LLM Inference Calculator · hn · 2026-08-28 · 6 upvotes · similarity 0.50
- PrunedCTC · github · 2026-09-27 · 40 upvotes · similarity 0.49
Other launches for this product
- No other launches for this product.