Hippius
mainsafetensorsbf16Public

happyconst/albedo-qwen3-4b-trainer-8

sha256:ca650df3946395e481a549d6482b1bb95a9718bf6c5b03562adedf5990b8342f·Indexed 17h ago

Parameters

4.0B

Total size

7.5 GB

Files

12

Quantization

BF16

No README on this version

Push a README.md with the model files to see it rendered here.

Model architecture

config.json
Architecture
Qwen3ForCausalLM
Model type
qwen3
Hidden size
2,560
Layers
36
Attention heads
32 (8 kv)
FFN size
9,728
Vocab size
151,936
Context window
40K
Dtype
bfloat16

Files

12 items
  • model-00002-of-00003.safetensors

    641ae7873eef

    3.7 GB

    safetensors

  • model-00001-of-00003.safetensors

    0698773bdf86

    3.7 GB

    safetensors

  • model-00003-of-00003.safetensors

    dd92da713d3c

    95.0 MB

    safetensors

  • tokenizer.json

    9aa803b36bea

    10.9 MB

  • vocab.json

    ca10d7e9fb3e

    2.6 MB

  • merges.txt

    8831e4f1a044

    1.6 MB

  • model.safetensors.index.json

    5b36dbd79cdd

    32.1 KB

  • tokenizer_config.json

    b142dbe443aa

    5.3 KB

  • config.json

    018398443a85

    727 B

  • added_tokens.json

    c0284b582e14

    707 B

  • special_tokens_map.json

    a36726f7fe39

    496 B

  • generation_config.json

    535ef5c50ad9

    214 B