Qwen/Qwen3.8-Flash-Next
Original177.4Bparameters
Fixture359.2Mparameters
Layer stacks2compared
Findings0to review
✓ Same architecture, smaller
Every module class, layer kind and checked property of the original is in the fixture.
Layer stacks
One square per layer, colored by kind: the layer's module classes and type attributes. Hover a square for its index and kind.
model.visual.blocks
Original · 27
Fixture · 27
| Kind | Original | Fixture | |
|---|---|---|---|
| VisionBlock | 27 | 27 | ✓ kept |
model.language_model.layers
Original · 48
Fixture · 4
| Kind | Original | Fixture | |
|---|---|---|---|
| linear_attention, Conv1d, TextGatedDeltaNet … | 35 | 2 | ✓ kept |
| linear_attention, Conv1d, Embedding … | 1 | 1 | ✓ kept |
| indexed_attention, TextAttention, TextQSAIndexer | 12 | 1 | ✓ kept |
Where the parameters are
Share of each model's parameters per component. Vocabulary and patch sizes are kept, so embeddings usually take a larger share of the fixture.
Original
Fixture
| Component | Original | Share | Fixture | Share |
|---|---|---|---|---|
| model.language_model.layers | 175.7B | 99.0% | 326.1M | 90.8% |
| model.language_model | 642.3M | 0.4% | 16.1M | 4.5% |
| lm_head | 635.7M | 0.4% | 15.9M | 4.4% |
| model.visual.blocks | 411.5M | 0.2% | 903.7K | 0.3% |
| model.visual | 37.5M | 0.0% | 328.2K | 0.1% |
Preserved properties
| Config | Property | Original | Fixture | |
|---|---|---|---|---|
| model | model_type | qwen4_exp | qwen4_exp | ✓ kept |
| model | tie_word_embeddings | False | False | ✓ kept |
| text_config | attention style | GQA (24 heads, 2 KV) | GQA (4 heads, 2 KV) | ✓ kept |
| text_config | MoE routing | top-10 of 512 experts | top-2 of 4 experts | ✓ kept |
| text_config | RoPE type | default | default | ✓ kept |
| text_config | model_type | qwen4_exp_text | qwen4_exp_text | ✓ kept |
| text_config | vocab_size | 248320 | 248320 | ✓ kept |
| text_config | tie_word_embeddings | False | False | ✓ kept |
| text_config | hidden_act | silu | silu | ✓ kept |
| text_config | max_position_embeddings | 262144 | 262144 | ✓ kept |
| vision_config | RoPE type | axial | axial | ✓ kept |
| vision_config | model_type | qwen4_exp_vision | qwen4_exp_vision | ✓ kept |
| vision_config | hidden_act | gelu_pytorch_tanh | gelu_pytorch_tanh | ✓ kept |
| vision_config | patch_size | 16 | 16 | ✓ kept |
Module classes
Instances of each class in the built model (prefix Qwen4Exp omitted).
| Class | Original | Fixture |
|---|---|---|
| Linear | 835 | 175 |
| TextRMSNorm | 148 | 16 |
| TextGatedResidual | 97 | 9 |
| SiLUActivation | 96 | 8 |
| LayerNorm | 55 | 55 |
| TextDecoderLayer | 48 | 4 |
| TextExperts | 48 | 4 |
| TextMLP | 48 | 4 |
| TextSparseMoeBlock | 48 | 4 |
| TextTopKRouter | 48 | 4 |
| Conv1d | 37 | 4 |
| TextGatedDeltaNet | 36 | 3 |
| TextRMSNormGated | 36 | 3 |
| GELUTanh | 27 | 27 |
| VisionAttention | 27 | 27 |
| VisionBlock | 27 | 27 |
| VisionMLP | 27 | 27 |
| TextAttention | 12 | 1 |
| TextQSAIndexer | 12 | 1 |
| Embedding | 3 | 3 |
| ModuleList | 2 | 2 |
| Conv3d | 1 | 1 |
| GELU | 1 | 1 |
| ForConditionalGeneration | 1 | 1 |
| Model | 1 | 1 |
| TextModel | 1 | 1 |
| TextNGramEmbedding | 1 | 1 |
| TextPLELayer | 1 | 1 |
| TextRotaryEmbedding | 1 | 1 |
| VisionModel | 1 | 1 |
| VisionPatchEmbed | 1 | 1 |
| VisionPatchMerger | 1 | 1 |
| VisionRotaryEmbedding | 1 | 1 |
Config changes
15 fields changed, 86 unchanged.
| Field | Original | Fixture |
|---|---|---|
| text_config.head_dim | 256 | 16 |
| text_config.hidden_size | 2560 | 64 |
| text_config.layer_types | 48 entries: 36× linear_attention, 12× indexed_attention | ["linear_attention", "linear_attention", "linear_attention", "indexed_attenti… |
| text_config.moe_intermediate_size | 640 | 32 |
| text_config.num_attention_heads | 24 | 4 |
| text_config.num_experts | 512 | 4 |
| text_config.num_experts_per_tok | 10 | 2 |
| text_config.num_hidden_layers | 48 | 4 |
| text_config.ple_embed_dim | 2560 | 16 |
| text_config.rope_parameters | {"mrope_interleaved": true, "mrope_section": [11, 11, 10], "partial_rotary_fa… | {"mrope_interleaved": true, "mrope_section": [3, 3, 2], "partial_rotary_facto… |
| text_config.shared_expert_intermediate_size | 640 | 128 |
| vision_config.hidden_size | 1152 | 64 |
| vision_config.intermediate_size | 4304 | 128 |
| vision_config.num_heads | 16 | 4 |
| vision_config.out_hidden_size | 2560 | 64 |