Dev Runs Uncensored DeepSeek Locally Using Dual Sparks at 40 tok/s
PMinervini · x · 2026-08-09
Developer Teknium shared their local deployment experience: by connecting two Spark units with a single cable, they ran an 'abliterated' (uncensored) DeepSeek v4 flash 0731 model without dspark, achieving completely private inference at around 40 tokens/s.
Other developers subsequently inquired about the exact method used for abliteration, speculating whether it utilized ds4's direction steering feature.
Related event: Dual Spark Runs Abliterated DeepSeek Locally at 40 tok/s(3 posts)→
More from Infra
- Amazon Proposes Per-Request Validated Evidence Caching for Production RAG — _reachsumit · 2026-10-16
- LinkedIn's TransX report: sequence transduction for recsys, +6.02% CTR in A/B tests — _reachsumit · 2026-10-16
- llama.cpp PR boosts CPU inference speed with merged AVX2 tiled accumulators — jacek2023 · 2026-10-16
- Zero's next-gen IVM engine runs in a single DO, live-querying a 100GiB R2 database — threepointone · 2026-10-16
- Meme: driving to the chip factory in 2028 captures AI's compute anxiety — beffjezos · 2026-10-16
- One volunteer maintains the time zone file behind 4 billion phones — danbri · 2026-10-16