CATALOGUE RECORD / HUB-DERIVED
Qwen
Qwen3-30B-A3B-FP8
Revision d206ba732169f29bb77fbf80fc2c4b81d4d30782
TENSOR ACCOUNTING
Where the bytes are.
Summed from the safetensors index, one row per dtype. A parameter count alone cannot produce this figure, because a checkpoint may mix widths.
| Dtype | Parameters | Bytes each | Bytes | Share |
|---|---|---|---|---|
F8_E4M3 | 29,896,998,912 | 1 | 29.9 GB | 92.1% |
F32 | 636,948,480 | 4 | 2.5 GB | 7.9% |
EVERY FIELD, WITH ITS ORIGIN
Sourced or undetermined. Never assumed.
Each value below names the exact API field it was computed from. Where the Hub does not establish a value, the reason is shown instead of a plausible default.
- Architecture
- qwen3_moe
config.model_type - Model classes
- Qwen3MoeForCausalLM
config.architectures - Routed experts
- 128
config.num_experts - Experts per token
- 8
config.num_experts_per_tok - Shared experts
- UndeterminedThe published config declares no always-on shared experts.
- Routing sparsity
- 16.0× (1 of every 16.0 experts)
config.num_experts / config.num_experts_per_tok - Total parameters
- 30,533,947,392
safetensors.total - Checkpoint bytes
- 32,444,792,832 (32.4 GB)
safetensors.parameters - Ships below 16-bit
- Yes
safetensors.parameters - Quantisation method
- fp8
config.quantization_config.quant_method - Trained context
- UndeterminedThe config summary omits max_position_embeddings. Trained context is a model-card claim, not a derivable fact.
- Declared licence
- apache-2.0
cardData.license - Base model
- Qwen/Qwen3-30B-A3B
cardData.base_model - Library
- transformers
library_name - Files in repository
- 17
siblings - Last modified
- 2025-07-26
lastModified
LICENCE POSTURE
Permissive
The repository declares a licence that is generally read as permitting commercial use. Read the licence file in the repository before relying on that.
This is a reading of a metadata field, not legal advice, and it does not account for the licences of upstream models or training data.
WEIGHT RESIDENCY FLOOR
The count below which it cannot fit.
Ceiling of checkpoint bytes over advertised accelerator memory. This is a lower bound on accelerator count for weights alone — KV cache, activations and runtime overhead all sit on top, so a real deployment needs more.