LFM3-1.7B-Web-ONNX / tokenizer.json

Commit History

9 days ago

  1. Block-32 GPTQ with a 48k vocabulary (900 MB of weights)

    Verified signature
    2395323
  2. Trim the vocabulary to 80k tokens (856 MB of weights)

    Verified signature
    c7cce6b
  3. LFM3-1.7B ONNX, GPTQ 4-bit (block 128), 16k rotary caches

    Verified signature
    5bc567e