Issues

• Native 1.0.12 grammar vocabulary lookup misses llm_config for Nemotron Super VL

  ashhart/TensorFold#951 · Filter by author swbiggart opened on Oct 16, 2026

• RFC: Nemotron 3.5 Bolt (nemotron_h_puzzle) on CUDA, and Nemotron Omni video

  ashhart/TensorFold#937 · Filter by author MiaAI-Lab opened on Oct 16, 2026

• tensorfold benchmark refuses Nemotron 3.5 Super VL: config.json over the 4 MiB read cap reports "Checkpoint is incomplete"

  ashhart/TensorFold#934 · Filter by author jkuepker opened on Oct 16, 2026

• Qwen3.6-35B-A3B (qwen3_5_moe) on the native Metal engine: proposal and gate

  ashhart/TensorFold#931 · Filter by author cshintov opened on Oct 16, 2026

• Native 27B on M3 Ultra: cold prefill still 1.7x the python engine (open since 1.0.3); uniform-4bit checkpoints bypass 1.0.10's faster tiles

  ashhart/TensorFold#918 · Filter by author astro-15-alive opened on Oct 16, 2026
  1

• Drama 3 (fishaudio/drama-3) text-to-speech on CUDA: family proposal and gate

  ashhart/TensorFold#914 · Filter by author plotarmordev opened on Oct 16, 2026
  1

• App 1.0.0-beta.6: notes from a first run against a remote OpenAI-compatible GLM server

  ashhart/TensorFold#884 · Filter by author Gogo6969 opened on Oct 15, 2026
  3

• GLM-5.3-Flash on one Mac: a short request waits until another stream's prompt is read, even with free streams

  ashhart/TensorFold#880 · Filter by author Gogo6969 opened on Oct 15, 2026
  1

• Living Weights: a later lesson's quiet refit can silence an earlier fact on its own questions, and nothing asks it again

  ashhart/TensorFold#870 · Filter by author ZackSample opened on Oct 15, 2026
  3

• Ling-3.1-flash on CUDA: a shared KDA kernel first, then the family at TP=1 and TP=2 (EXL3)

  ashhart/TensorFold#850 · Filter by author jschmied opened on Oct 14, 2026
  5

• DeepSeek-V4.1 on the native engine: one family, CUDA and Metal (tracker)

  ashhart/TensorFold#847 · Filter by author ashhart opened on Oct 14, 2026
  7

• Python engine features to carry into the native engine

  ashhart/TensorFold#840 · Filter by author ashhart opened on Oct 14, 2026
  3

• EXL3 checkpoints on the native engine (tracker)

  ashhart/TensorFold#832 · Filter by author ashhart opened on Oct 14, 2026
  2

• Two-Mac (--speed-up) for Qwen3.8-27B, and a mid-size model that fits a pair of 128 GB Macs

  ashhart/TensorFold#808 · Filter by author appljd opened on Oct 13, 2026
  1

• qwen3_5_moe on native CUDA with 3-bit routed experts (a 35B-A3B on a 24 GB card): where it lives, and its gate

  ashhart/TensorFold#803 · Filter by author CanadaApollo6 opened on Oct 13, 2026
  5

• qwen27 CUDA: an opt-in 4-bit attention cache (262K on a 24 GB card) — agreeing on its gate

  ashhart/TensorFold#799 · Filter by author CanadaApollo6 opened on Oct 13, 2026
  2

• Qwen3.8-27B + DFlash2 with 4 concurrent sampled requests: native 1.0.8 yields 2.9 tokens/round against 0.6.5's 5.4, so a batch Q&A job takes 46% longer

  ashhart/TensorFold#790 · Filter by author appljd opened on Oct 13, 2026
  2

• New family: MiniCPM-V 4.7 35B-A3B (Qwen3.6 MoE fine-tune + vision tower)

  ashhart/TensorFold#785 · Filter by author jkuepker opened on Oct 13, 2026
  5

• Universal model support, Living Weights with learn/unlearn, consumer GPU support, and Codex CLI compatibility

  ashhart/TensorFold#782 · Filter by author XAEAZii opened on Oct 13, 2026
  2

• Proposal: streams take memory from one core pool as they grow, instead of a full window each at load

  ashhart/TensorFold#781 · Filter by author jschmied opened on Oct 13, 2026
  7

• Feature request: a llama.cpp-compatible GET /slots (per-stream state) — /health says how many wait, not why

  ashhart/TensorFold#752 · Filter by author ThefloorMiner opened on Oct 12, 2026
  1

• Python server: default --max-tokens 4096 counts thinking tokens — "unlimited" thinking ends with finish_reason=length and an empty reply

  ashhart/TensorFold#749 · Filter by author ThefloorMiner opened on Oct 12, 2026
  1

• Metal (MLX): loading 180 GB of weights from an SMB share stalls mx.eval past the Metal watchdog — GPU Timeout; the same files load from local SSD in under a minute

  ashhart/TensorFold#748 · Filter by author ThefloorMiner opened on Oct 12, 2026
  1

• Proposal: restore single-Mac GLM-5.3-Flash on the M3 Ultra

  ashhart/TensorFold#743 · Filter by author harrisonfriia opened on Oct 12, 2026
  3

• Proposal: MiMo-V2.6-Flash on the CUDA native engine across two GPUs (EXL3, DFlash)

  ashhart/TensorFold#725 · Filter by author ajensenwaud opened on Oct 12, 2026
  3