Upload folder using huggingface_hub
Browse files- README.md +25 -0
- assets/tokenizer/merges.txt +0 -0
- assets/tokenizer/vocabulary.json +0 -0
- config.json +22 -0
- metadata.json +9 -0
- model.weights.json +407 -0
- model_00000.weights.h5 +3 -0
- model_00001.weights.h5 +3 -0
- model_00002.weights.h5 +3 -0
- model_00003.weights.h5 +3 -0
- preprocessor.json +69 -0
- task.json +99 -0
- tokenizer.json +48 -0
README.md
ADDED
|
@@ -0,0 +1,25 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: keras-hub
|
| 3 |
+
pipeline_tag: text-generation
|
| 4 |
+
---
|
| 5 |
+
This is a [`Qwen3` model](https://keras.io/api/keras_hub/models/qwen3) uploaded using the KerasHub library and can be used with JAX, TensorFlow, and PyTorch backends.
|
| 6 |
+
This model is related to a `CausalLM` task.
|
| 7 |
+
|
| 8 |
+
Model config:
|
| 9 |
+
* **name:** qwen3_backbone
|
| 10 |
+
* **trainable:** True
|
| 11 |
+
* **vocabulary_size:** 151936
|
| 12 |
+
* **num_layers:** 36
|
| 13 |
+
* **num_query_heads:** 32
|
| 14 |
+
* **hidden_dim:** 4096
|
| 15 |
+
* **head_dim:** 128
|
| 16 |
+
* **intermediate_dim:** 12288
|
| 17 |
+
* **rope_max_wavelength:** 1000000
|
| 18 |
+
* **rope_scaling_factor:** 1.0
|
| 19 |
+
* **num_key_value_heads:** 8
|
| 20 |
+
* **layer_norm_epsilon:** 1e-06
|
| 21 |
+
* **dropout:** 0.0
|
| 22 |
+
* **tie_word_embeddings:** False
|
| 23 |
+
* **sliding_window_size:** None
|
| 24 |
+
|
| 25 |
+
This model card has been generated automatically and should be completed by the model author. See [Model Cards documentation](https://huggingface.co/docs/hub/model-cards) for more information.
|
assets/tokenizer/merges.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
assets/tokenizer/vocabulary.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
config.json
ADDED
|
@@ -0,0 +1,22 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"module": "keras_hub.src.models.qwen3.qwen3_backbone",
|
| 3 |
+
"class_name": "Qwen3Backbone",
|
| 4 |
+
"config": {
|
| 5 |
+
"name": "qwen3_backbone",
|
| 6 |
+
"trainable": true,
|
| 7 |
+
"vocabulary_size": 151936,
|
| 8 |
+
"num_layers": 36,
|
| 9 |
+
"num_query_heads": 32,
|
| 10 |
+
"hidden_dim": 4096,
|
| 11 |
+
"head_dim": 128,
|
| 12 |
+
"intermediate_dim": 12288,
|
| 13 |
+
"rope_max_wavelength": 1000000,
|
| 14 |
+
"rope_scaling_factor": 1.0,
|
| 15 |
+
"num_key_value_heads": 8,
|
| 16 |
+
"layer_norm_epsilon": 1e-06,
|
| 17 |
+
"dropout": 0.0,
|
| 18 |
+
"tie_word_embeddings": false,
|
| 19 |
+
"sliding_window_size": null
|
| 20 |
+
},
|
| 21 |
+
"registered_name": "keras_hub>Qwen3Backbone"
|
| 22 |
+
}
|
metadata.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"keras_version": "3.10.0.dev2025071003",
|
| 3 |
+
"keras_hub_version": "0.22.0.dev0",
|
| 4 |
+
"parameter_count": 8190735360,
|
| 5 |
+
"date_saved": "2025-07-10@20:20:38",
|
| 6 |
+
"tasks": [
|
| 7 |
+
"CausalLM"
|
| 8 |
+
]
|
| 9 |
+
}
|
model.weights.json
ADDED
|
@@ -0,0 +1,407 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"metadata": {
|
| 3 |
+
"total_size": 32762941440.0
|
| 4 |
+
},
|
| 5 |
+
"weight_map": {
|
| 6 |
+
"/layers/reversible_embedding/vars": [
|
| 7 |
+
"model_00000.weights.h5"
|
| 8 |
+
],
|
| 9 |
+
"/layers/qwen3_transformer_decoder/_feedforward_gate_dense/vars": "model_00000.weights.h5",
|
| 10 |
+
"/layers/qwen3_transformer_decoder/_feedforward_intermediate_dense/vars": "model_00000.weights.h5",
|
| 11 |
+
"/layers/qwen3_transformer_decoder/_feedforward_layernorm/vars": "model_00000.weights.h5",
|
| 12 |
+
"/layers/qwen3_transformer_decoder/_feedforward_output_dense/vars": "model_00000.weights.h5",
|
| 13 |
+
"/layers/qwen3_transformer_decoder/_self_attention_layer/_key_dense/vars": "model_00000.weights.h5",
|
| 14 |
+
"/layers/qwen3_transformer_decoder/_self_attention_layer/_key_dense_layer_norm/vars": "model_00000.weights.h5",
|
| 15 |
+
"/layers/qwen3_transformer_decoder/_self_attention_layer/_output_dense/vars": "model_00000.weights.h5",
|
| 16 |
+
"/layers/qwen3_transformer_decoder/_self_attention_layer/_query_dense/vars": "model_00000.weights.h5",
|
| 17 |
+
"/layers/qwen3_transformer_decoder/_self_attention_layer/_query_dense_layer_norm/vars": "model_00000.weights.h5",
|
| 18 |
+
"/layers/qwen3_transformer_decoder/_self_attention_layer/_value_dense/vars": "model_00000.weights.h5",
|
| 19 |
+
"/layers/qwen3_transformer_decoder/_self_attention_layernorm/vars": "model_00000.weights.h5",
|
| 20 |
+
"/layers/qwen3_transformer_decoder_1/_feedforward_gate_dense/vars": "model_00000.weights.h5",
|
| 21 |
+
"/layers/qwen3_transformer_decoder_1/_feedforward_intermediate_dense/vars": "model_00000.weights.h5",
|
| 22 |
+
"/layers/qwen3_transformer_decoder_1/_feedforward_layernorm/vars": "model_00000.weights.h5",
|
| 23 |
+
"/layers/qwen3_transformer_decoder_1/_feedforward_output_dense/vars": "model_00000.weights.h5",
|
| 24 |
+
"/layers/qwen3_transformer_decoder_1/_self_attention_layer/_key_dense/vars": "model_00000.weights.h5",
|
| 25 |
+
"/layers/qwen3_transformer_decoder_1/_self_attention_layer/_key_dense_layer_norm/vars": "model_00000.weights.h5",
|
| 26 |
+
"/layers/qwen3_transformer_decoder_1/_self_attention_layer/_output_dense/vars": "model_00000.weights.h5",
|
| 27 |
+
"/layers/qwen3_transformer_decoder_1/_self_attention_layer/_query_dense/vars": "model_00000.weights.h5",
|
| 28 |
+
"/layers/qwen3_transformer_decoder_1/_self_attention_layer/_query_dense_layer_norm/vars": "model_00000.weights.h5",
|
| 29 |
+
"/layers/qwen3_transformer_decoder_1/_self_attention_layer/_value_dense/vars": "model_00000.weights.h5",
|
| 30 |
+
"/layers/qwen3_transformer_decoder_1/_self_attention_layernorm/vars": "model_00000.weights.h5",
|
| 31 |
+
"/layers/qwen3_transformer_decoder_2/_feedforward_gate_dense/vars": "model_00000.weights.h5",
|
| 32 |
+
"/layers/qwen3_transformer_decoder_2/_feedforward_intermediate_dense/vars": "model_00000.weights.h5",
|
| 33 |
+
"/layers/qwen3_transformer_decoder_2/_feedforward_layernorm/vars": "model_00000.weights.h5",
|
| 34 |
+
"/layers/qwen3_transformer_decoder_2/_feedforward_output_dense/vars": "model_00000.weights.h5",
|
| 35 |
+
"/layers/qwen3_transformer_decoder_2/_self_attention_layer/_key_dense/vars": "model_00000.weights.h5",
|
| 36 |
+
"/layers/qwen3_transformer_decoder_2/_self_attention_layer/_key_dense_layer_norm/vars": "model_00000.weights.h5",
|
| 37 |
+
"/layers/qwen3_transformer_decoder_2/_self_attention_layer/_output_dense/vars": "model_00000.weights.h5",
|
| 38 |
+
"/layers/qwen3_transformer_decoder_2/_self_attention_layer/_query_dense/vars": "model_00000.weights.h5",
|
| 39 |
+
"/layers/qwen3_transformer_decoder_2/_self_attention_layer/_query_dense_layer_norm/vars": "model_00000.weights.h5",
|
| 40 |
+
"/layers/qwen3_transformer_decoder_2/_self_attention_layer/_value_dense/vars": "model_00000.weights.h5",
|
| 41 |
+
"/layers/qwen3_transformer_decoder_2/_self_attention_layernorm/vars": "model_00000.weights.h5",
|
| 42 |
+
"/layers/qwen3_transformer_decoder_3/_feedforward_gate_dense/vars": "model_00000.weights.h5",
|
| 43 |
+
"/layers/qwen3_transformer_decoder_3/_feedforward_intermediate_dense/vars": "model_00000.weights.h5",
|
| 44 |
+
"/layers/qwen3_transformer_decoder_3/_feedforward_layernorm/vars": "model_00000.weights.h5",
|
| 45 |
+
"/layers/qwen3_transformer_decoder_3/_feedforward_output_dense/vars": "model_00000.weights.h5",
|
| 46 |
+
"/layers/qwen3_transformer_decoder_3/_self_attention_layer/_key_dense/vars": "model_00000.weights.h5",
|
| 47 |
+
"/layers/qwen3_transformer_decoder_3/_self_attention_layer/_key_dense_layer_norm/vars": "model_00000.weights.h5",
|
| 48 |
+
"/layers/qwen3_transformer_decoder_3/_self_attention_layer/_output_dense/vars": "model_00000.weights.h5",
|
| 49 |
+
"/layers/qwen3_transformer_decoder_3/_self_attention_layer/_query_dense/vars": "model_00000.weights.h5",
|
| 50 |
+
"/layers/qwen3_transformer_decoder_3/_self_attention_layer/_query_dense_layer_norm/vars": "model_00000.weights.h5",
|
| 51 |
+
"/layers/qwen3_transformer_decoder_3/_self_attention_layer/_value_dense/vars": "model_00000.weights.h5",
|
| 52 |
+
"/layers/qwen3_transformer_decoder_3/_self_attention_layernorm/vars": "model_00000.weights.h5",
|
| 53 |
+
"/layers/qwen3_transformer_decoder_4/_feedforward_gate_dense/vars": "model_00000.weights.h5",
|
| 54 |
+
"/layers/qwen3_transformer_decoder_4/_feedforward_intermediate_dense/vars": "model_00000.weights.h5",
|
| 55 |
+
"/layers/qwen3_transformer_decoder_4/_feedforward_layernorm/vars": "model_00000.weights.h5",
|
| 56 |
+
"/layers/qwen3_transformer_decoder_4/_feedforward_output_dense/vars": "model_00000.weights.h5",
|
| 57 |
+
"/layers/qwen3_transformer_decoder_4/_self_attention_layer/_key_dense/vars": "model_00000.weights.h5",
|
| 58 |
+
"/layers/qwen3_transformer_decoder_4/_self_attention_layer/_key_dense_layer_norm/vars": "model_00000.weights.h5",
|
| 59 |
+
"/layers/qwen3_transformer_decoder_4/_self_attention_layer/_output_dense/vars": "model_00000.weights.h5",
|
| 60 |
+
"/layers/qwen3_transformer_decoder_4/_self_attention_layer/_query_dense/vars": "model_00000.weights.h5",
|
| 61 |
+
"/layers/qwen3_transformer_decoder_4/_self_attention_layer/_query_dense_layer_norm/vars": "model_00000.weights.h5",
|
| 62 |
+
"/layers/qwen3_transformer_decoder_4/_self_attention_layer/_value_dense/vars": "model_00000.weights.h5",
|
| 63 |
+
"/layers/qwen3_transformer_decoder_4/_self_attention_layernorm/vars": "model_00000.weights.h5",
|
| 64 |
+
"/layers/qwen3_transformer_decoder_5/_feedforward_gate_dense/vars": "model_00000.weights.h5",
|
| 65 |
+
"/layers/qwen3_transformer_decoder_5/_feedforward_intermediate_dense/vars": "model_00000.weights.h5",
|
| 66 |
+
"/layers/qwen3_transformer_decoder_5/_feedforward_layernorm/vars": "model_00000.weights.h5",
|
| 67 |
+
"/layers/qwen3_transformer_decoder_5/_feedforward_output_dense/vars": "model_00000.weights.h5",
|
| 68 |
+
"/layers/qwen3_transformer_decoder_5/_self_attention_layer/_key_dense/vars": "model_00000.weights.h5",
|
| 69 |
+
"/layers/qwen3_transformer_decoder_5/_self_attention_layer/_key_dense_layer_norm/vars": "model_00000.weights.h5",
|
| 70 |
+
"/layers/qwen3_transformer_decoder_5/_self_attention_layer/_output_dense/vars": "model_00000.weights.h5",
|
| 71 |
+
"/layers/qwen3_transformer_decoder_5/_self_attention_layer/_query_dense/vars": "model_00000.weights.h5",
|
| 72 |
+
"/layers/qwen3_transformer_decoder_5/_self_attention_layer/_query_dense_layer_norm/vars": "model_00000.weights.h5",
|
| 73 |
+
"/layers/qwen3_transformer_decoder_5/_self_attention_layer/_value_dense/vars": "model_00000.weights.h5",
|
| 74 |
+
"/layers/qwen3_transformer_decoder_5/_self_attention_layernorm/vars": "model_00000.weights.h5",
|
| 75 |
+
"/layers/qwen3_transformer_decoder_6/_feedforward_gate_dense/vars": "model_00000.weights.h5",
|
| 76 |
+
"/layers/qwen3_transformer_decoder_6/_feedforward_intermediate_dense/vars": "model_00000.weights.h5",
|
| 77 |
+
"/layers/qwen3_transformer_decoder_6/_feedforward_layernorm/vars": "model_00000.weights.h5",
|
| 78 |
+
"/layers/qwen3_transformer_decoder_6/_feedforward_output_dense/vars": "model_00000.weights.h5",
|
| 79 |
+
"/layers/qwen3_transformer_decoder_6/_self_attention_layer/_key_dense/vars": "model_00000.weights.h5",
|
| 80 |
+
"/layers/qwen3_transformer_decoder_6/_self_attention_layer/_key_dense_layer_norm/vars": "model_00000.weights.h5",
|
| 81 |
+
"/layers/qwen3_transformer_decoder_6/_self_attention_layer/_output_dense/vars": "model_00000.weights.h5",
|
| 82 |
+
"/layers/qwen3_transformer_decoder_6/_self_attention_layer/_query_dense/vars": "model_00000.weights.h5",
|
| 83 |
+
"/layers/qwen3_transformer_decoder_6/_self_attention_layer/_query_dense_layer_norm/vars": "model_00000.weights.h5",
|
| 84 |
+
"/layers/qwen3_transformer_decoder_6/_self_attention_layer/_value_dense/vars": "model_00000.weights.h5",
|
| 85 |
+
"/layers/qwen3_transformer_decoder_6/_self_attention_layernorm/vars": "model_00000.weights.h5",
|
| 86 |
+
"/layers/qwen3_transformer_decoder_7/_feedforward_gate_dense/vars": "model_00000.weights.h5",
|
| 87 |
+
"/layers/qwen3_transformer_decoder_7/_feedforward_intermediate_dense/vars": "model_00001.weights.h5",
|
| 88 |
+
"/layers/qwen3_transformer_decoder_7/_feedforward_layernorm/vars": "model_00001.weights.h5",
|
| 89 |
+
"/layers/qwen3_transformer_decoder_7/_feedforward_output_dense/vars": "model_00001.weights.h5",
|
| 90 |
+
"/layers/qwen3_transformer_decoder_7/_self_attention_layer/_key_dense/vars": "model_00001.weights.h5",
|
| 91 |
+
"/layers/qwen3_transformer_decoder_7/_self_attention_layer/_key_dense_layer_norm/vars": "model_00001.weights.h5",
|
| 92 |
+
"/layers/qwen3_transformer_decoder_7/_self_attention_layer/_output_dense/vars": "model_00001.weights.h5",
|
| 93 |
+
"/layers/qwen3_transformer_decoder_7/_self_attention_layer/_query_dense/vars": "model_00001.weights.h5",
|
| 94 |
+
"/layers/qwen3_transformer_decoder_7/_self_attention_layer/_query_dense_layer_norm/vars": "model_00001.weights.h5",
|
| 95 |
+
"/layers/qwen3_transformer_decoder_7/_self_attention_layer/_value_dense/vars": "model_00001.weights.h5",
|
| 96 |
+
"/layers/qwen3_transformer_decoder_7/_self_attention_layernorm/vars": "model_00001.weights.h5",
|
| 97 |
+
"/layers/qwen3_transformer_decoder_8/_feedforward_gate_dense/vars": "model_00001.weights.h5",
|
| 98 |
+
"/layers/qwen3_transformer_decoder_8/_feedforward_intermediate_dense/vars": "model_00001.weights.h5",
|
| 99 |
+
"/layers/qwen3_transformer_decoder_8/_feedforward_layernorm/vars": "model_00001.weights.h5",
|
| 100 |
+
"/layers/qwen3_transformer_decoder_8/_feedforward_output_dense/vars": "model_00001.weights.h5",
|
| 101 |
+
"/layers/qwen3_transformer_decoder_8/_self_attention_layer/_key_dense/vars": "model_00001.weights.h5",
|
| 102 |
+
"/layers/qwen3_transformer_decoder_8/_self_attention_layer/_key_dense_layer_norm/vars": "model_00001.weights.h5",
|
| 103 |
+
"/layers/qwen3_transformer_decoder_8/_self_attention_layer/_output_dense/vars": "model_00001.weights.h5",
|
| 104 |
+
"/layers/qwen3_transformer_decoder_8/_self_attention_layer/_query_dense/vars": "model_00001.weights.h5",
|
| 105 |
+
"/layers/qwen3_transformer_decoder_8/_self_attention_layer/_query_dense_layer_norm/vars": "model_00001.weights.h5",
|
| 106 |
+
"/layers/qwen3_transformer_decoder_8/_self_attention_layer/_value_dense/vars": "model_00001.weights.h5",
|
| 107 |
+
"/layers/qwen3_transformer_decoder_8/_self_attention_layernorm/vars": "model_00001.weights.h5",
|
| 108 |
+
"/layers/qwen3_transformer_decoder_9/_feedforward_gate_dense/vars": "model_00001.weights.h5",
|
| 109 |
+
"/layers/qwen3_transformer_decoder_9/_feedforward_intermediate_dense/vars": "model_00001.weights.h5",
|
| 110 |
+
"/layers/qwen3_transformer_decoder_9/_feedforward_layernorm/vars": "model_00001.weights.h5",
|
| 111 |
+
"/layers/qwen3_transformer_decoder_9/_feedforward_output_dense/vars": "model_00001.weights.h5",
|
| 112 |
+
"/layers/qwen3_transformer_decoder_9/_self_attention_layer/_key_dense/vars": "model_00001.weights.h5",
|
| 113 |
+
"/layers/qwen3_transformer_decoder_9/_self_attention_layer/_key_dense_layer_norm/vars": "model_00001.weights.h5",
|
| 114 |
+
"/layers/qwen3_transformer_decoder_9/_self_attention_layer/_output_dense/vars": "model_00001.weights.h5",
|
| 115 |
+
"/layers/qwen3_transformer_decoder_9/_self_attention_layer/_query_dense/vars": "model_00001.weights.h5",
|
| 116 |
+
"/layers/qwen3_transformer_decoder_9/_self_attention_layer/_query_dense_layer_norm/vars": "model_00001.weights.h5",
|
| 117 |
+
"/layers/qwen3_transformer_decoder_9/_self_attention_layer/_value_dense/vars": "model_00001.weights.h5",
|
| 118 |
+
"/layers/qwen3_transformer_decoder_9/_self_attention_layernorm/vars": "model_00001.weights.h5",
|
| 119 |
+
"/layers/qwen3_transformer_decoder_10/_feedforward_gate_dense/vars": "model_00001.weights.h5",
|
| 120 |
+
"/layers/qwen3_transformer_decoder_10/_feedforward_intermediate_dense/vars": "model_00001.weights.h5",
|
| 121 |
+
"/layers/qwen3_transformer_decoder_10/_feedforward_layernorm/vars": "model_00001.weights.h5",
|
| 122 |
+
"/layers/qwen3_transformer_decoder_10/_feedforward_output_dense/vars": "model_00001.weights.h5",
|
| 123 |
+
"/layers/qwen3_transformer_decoder_10/_self_attention_layer/_key_dense/vars": "model_00001.weights.h5",
|
| 124 |
+
"/layers/qwen3_transformer_decoder_10/_self_attention_layer/_key_dense_layer_norm/vars": "model_00001.weights.h5",
|
| 125 |
+
"/layers/qwen3_transformer_decoder_10/_self_attention_layer/_output_dense/vars": "model_00001.weights.h5",
|
| 126 |
+
"/layers/qwen3_transformer_decoder_10/_self_attention_layer/_query_dense/vars": "model_00001.weights.h5",
|
| 127 |
+
"/layers/qwen3_transformer_decoder_10/_self_attention_layer/_query_dense_layer_norm/vars": "model_00001.weights.h5",
|
| 128 |
+
"/layers/qwen3_transformer_decoder_10/_self_attention_layer/_value_dense/vars": "model_00001.weights.h5",
|
| 129 |
+
"/layers/qwen3_transformer_decoder_10/_self_attention_layernorm/vars": "model_00001.weights.h5",
|
| 130 |
+
"/layers/qwen3_transformer_decoder_11/_feedforward_gate_dense/vars": "model_00001.weights.h5",
|
| 131 |
+
"/layers/qwen3_transformer_decoder_11/_feedforward_intermediate_dense/vars": "model_00001.weights.h5",
|
| 132 |
+
"/layers/qwen3_transformer_decoder_11/_feedforward_layernorm/vars": "model_00001.weights.h5",
|
| 133 |
+
"/layers/qwen3_transformer_decoder_11/_feedforward_output_dense/vars": "model_00001.weights.h5",
|
| 134 |
+
"/layers/qwen3_transformer_decoder_11/_self_attention_layer/_key_dense/vars": "model_00001.weights.h5",
|
| 135 |
+
"/layers/qwen3_transformer_decoder_11/_self_attention_layer/_key_dense_layer_norm/vars": "model_00001.weights.h5",
|
| 136 |
+
"/layers/qwen3_transformer_decoder_11/_self_attention_layer/_output_dense/vars": "model_00001.weights.h5",
|
| 137 |
+
"/layers/qwen3_transformer_decoder_11/_self_attention_layer/_query_dense/vars": "model_00001.weights.h5",
|
| 138 |
+
"/layers/qwen3_transformer_decoder_11/_self_attention_layer/_query_dense_layer_norm/vars": "model_00001.weights.h5",
|
| 139 |
+
"/layers/qwen3_transformer_decoder_11/_self_attention_layer/_value_dense/vars": "model_00001.weights.h5",
|
| 140 |
+
"/layers/qwen3_transformer_decoder_11/_self_attention_layernorm/vars": "model_00001.weights.h5",
|
| 141 |
+
"/layers/qwen3_transformer_decoder_12/_feedforward_gate_dense/vars": "model_00001.weights.h5",
|
| 142 |
+
"/layers/qwen3_transformer_decoder_12/_feedforward_intermediate_dense/vars": "model_00001.weights.h5",
|
| 143 |
+
"/layers/qwen3_transformer_decoder_12/_feedforward_layernorm/vars": "model_00001.weights.h5",
|
| 144 |
+
"/layers/qwen3_transformer_decoder_12/_feedforward_output_dense/vars": "model_00001.weights.h5",
|
| 145 |
+
"/layers/qwen3_transformer_decoder_12/_self_attention_layer/_key_dense/vars": "model_00001.weights.h5",
|
| 146 |
+
"/layers/qwen3_transformer_decoder_12/_self_attention_layer/_key_dense_layer_norm/vars": "model_00001.weights.h5",
|
| 147 |
+
"/layers/qwen3_transformer_decoder_12/_self_attention_layer/_output_dense/vars": "model_00001.weights.h5",
|
| 148 |
+
"/layers/qwen3_transformer_decoder_12/_self_attention_layer/_query_dense/vars": "model_00001.weights.h5",
|
| 149 |
+
"/layers/qwen3_transformer_decoder_12/_self_attention_layer/_query_dense_layer_norm/vars": "model_00001.weights.h5",
|
| 150 |
+
"/layers/qwen3_transformer_decoder_12/_self_attention_layer/_value_dense/vars": "model_00001.weights.h5",
|
| 151 |
+
"/layers/qwen3_transformer_decoder_12/_self_attention_layernorm/vars": "model_00001.weights.h5",
|
| 152 |
+
"/layers/qwen3_transformer_decoder_13/_feedforward_gate_dense/vars": "model_00001.weights.h5",
|
| 153 |
+
"/layers/qwen3_transformer_decoder_13/_feedforward_intermediate_dense/vars": "model_00001.weights.h5",
|
| 154 |
+
"/layers/qwen3_transformer_decoder_13/_feedforward_layernorm/vars": "model_00001.weights.h5",
|
| 155 |
+
"/layers/qwen3_transformer_decoder_13/_feedforward_output_dense/vars": "model_00001.weights.h5",
|
| 156 |
+
"/layers/qwen3_transformer_decoder_13/_self_attention_layer/_key_dense/vars": "model_00001.weights.h5",
|
| 157 |
+
"/layers/qwen3_transformer_decoder_13/_self_attention_layer/_key_dense_layer_norm/vars": "model_00001.weights.h5",
|
| 158 |
+
"/layers/qwen3_transformer_decoder_13/_self_attention_layer/_output_dense/vars": "model_00001.weights.h5",
|
| 159 |
+
"/layers/qwen3_transformer_decoder_13/_self_attention_layer/_query_dense/vars": "model_00001.weights.h5",
|
| 160 |
+
"/layers/qwen3_transformer_decoder_13/_self_attention_layer/_query_dense_layer_norm/vars": "model_00001.weights.h5",
|
| 161 |
+
"/layers/qwen3_transformer_decoder_13/_self_attention_layer/_value_dense/vars": "model_00001.weights.h5",
|
| 162 |
+
"/layers/qwen3_transformer_decoder_13/_self_attention_layernorm/vars": "model_00001.weights.h5",
|
| 163 |
+
"/layers/qwen3_transformer_decoder_14/_feedforward_gate_dense/vars": "model_00001.weights.h5",
|
| 164 |
+
"/layers/qwen3_transformer_decoder_14/_feedforward_intermediate_dense/vars": "model_00001.weights.h5",
|
| 165 |
+
"/layers/qwen3_transformer_decoder_14/_feedforward_layernorm/vars": "model_00001.weights.h5",
|
| 166 |
+
"/layers/qwen3_transformer_decoder_14/_feedforward_output_dense/vars": "model_00001.weights.h5",
|
| 167 |
+
"/layers/qwen3_transformer_decoder_14/_self_attention_layer/_key_dense/vars": "model_00001.weights.h5",
|
| 168 |
+
"/layers/qwen3_transformer_decoder_14/_self_attention_layer/_key_dense_layer_norm/vars": "model_00001.weights.h5",
|
| 169 |
+
"/layers/qwen3_transformer_decoder_14/_self_attention_layer/_output_dense/vars": "model_00001.weights.h5",
|
| 170 |
+
"/layers/qwen3_transformer_decoder_14/_self_attention_layer/_query_dense/vars": "model_00001.weights.h5",
|
| 171 |
+
"/layers/qwen3_transformer_decoder_14/_self_attention_layer/_query_dense_layer_norm/vars": "model_00001.weights.h5",
|
| 172 |
+
"/layers/qwen3_transformer_decoder_14/_self_attention_layer/_value_dense/vars": "model_00001.weights.h5",
|
| 173 |
+
"/layers/qwen3_transformer_decoder_14/_self_attention_layernorm/vars": "model_00001.weights.h5",
|
| 174 |
+
"/layers/qwen3_transformer_decoder_15/_feedforward_gate_dense/vars": "model_00001.weights.h5",
|
| 175 |
+
"/layers/qwen3_transformer_decoder_15/_feedforward_intermediate_dense/vars": "model_00001.weights.h5",
|
| 176 |
+
"/layers/qwen3_transformer_decoder_15/_feedforward_layernorm/vars": "model_00001.weights.h5",
|
| 177 |
+
"/layers/qwen3_transformer_decoder_15/_feedforward_output_dense/vars": "model_00001.weights.h5",
|
| 178 |
+
"/layers/qwen3_transformer_decoder_15/_self_attention_layer/_key_dense/vars": "model_00001.weights.h5",
|
| 179 |
+
"/layers/qwen3_transformer_decoder_15/_self_attention_layer/_key_dense_layer_norm/vars": "model_00001.weights.h5",
|
| 180 |
+
"/layers/qwen3_transformer_decoder_15/_self_attention_layer/_output_dense/vars": "model_00001.weights.h5",
|
| 181 |
+
"/layers/qwen3_transformer_decoder_15/_self_attention_layer/_query_dense/vars": "model_00001.weights.h5",
|
| 182 |
+
"/layers/qwen3_transformer_decoder_15/_self_attention_layer/_query_dense_layer_norm/vars": "model_00001.weights.h5",
|
| 183 |
+
"/layers/qwen3_transformer_decoder_15/_self_attention_layer/_value_dense/vars": "model_00001.weights.h5",
|
| 184 |
+
"/layers/qwen3_transformer_decoder_15/_self_attention_layernorm/vars": "model_00001.weights.h5",
|
| 185 |
+
"/layers/qwen3_transformer_decoder_16/_feedforward_gate_dense/vars": "model_00001.weights.h5",
|
| 186 |
+
"/layers/qwen3_transformer_decoder_16/_feedforward_intermediate_dense/vars": "model_00001.weights.h5",
|
| 187 |
+
"/layers/qwen3_transformer_decoder_16/_feedforward_layernorm/vars": "model_00001.weights.h5",
|
| 188 |
+
"/layers/qwen3_transformer_decoder_16/_feedforward_output_dense/vars": "model_00001.weights.h5",
|
| 189 |
+
"/layers/qwen3_transformer_decoder_16/_self_attention_layer/_key_dense/vars": "model_00001.weights.h5",
|
| 190 |
+
"/layers/qwen3_transformer_decoder_16/_self_attention_layer/_key_dense_layer_norm/vars": "model_00001.weights.h5",
|
| 191 |
+
"/layers/qwen3_transformer_decoder_16/_self_attention_layer/_output_dense/vars": "model_00001.weights.h5",
|
| 192 |
+
"/layers/qwen3_transformer_decoder_16/_self_attention_layer/_query_dense/vars": "model_00001.weights.h5",
|
| 193 |
+
"/layers/qwen3_transformer_decoder_16/_self_attention_layer/_query_dense_layer_norm/vars": "model_00001.weights.h5",
|
| 194 |
+
"/layers/qwen3_transformer_decoder_16/_self_attention_layer/_value_dense/vars": "model_00001.weights.h5",
|
| 195 |
+
"/layers/qwen3_transformer_decoder_16/_self_attention_layernorm/vars": "model_00001.weights.h5",
|
| 196 |
+
"/layers/qwen3_transformer_decoder_17/_feedforward_gate_dense/vars": "model_00001.weights.h5",
|
| 197 |
+
"/layers/qwen3_transformer_decoder_17/_feedforward_intermediate_dense/vars": "model_00001.weights.h5",
|
| 198 |
+
"/layers/qwen3_transformer_decoder_17/_feedforward_layernorm/vars": "model_00001.weights.h5",
|
| 199 |
+
"/layers/qwen3_transformer_decoder_17/_feedforward_output_dense/vars": "model_00001.weights.h5",
|
| 200 |
+
"/layers/qwen3_transformer_decoder_17/_self_attention_layer/_key_dense/vars": "model_00001.weights.h5",
|
| 201 |
+
"/layers/qwen3_transformer_decoder_17/_self_attention_layer/_key_dense_layer_norm/vars": "model_00001.weights.h5",
|
| 202 |
+
"/layers/qwen3_transformer_decoder_17/_self_attention_layer/_output_dense/vars": "model_00001.weights.h5",
|
| 203 |
+
"/layers/qwen3_transformer_decoder_17/_self_attention_layer/_query_dense/vars": "model_00001.weights.h5",
|
| 204 |
+
"/layers/qwen3_transformer_decoder_17/_self_attention_layer/_query_dense_layer_norm/vars": "model_00001.weights.h5",
|
| 205 |
+
"/layers/qwen3_transformer_decoder_17/_self_attention_layer/_value_dense/vars": "model_00001.weights.h5",
|
| 206 |
+
"/layers/qwen3_transformer_decoder_17/_self_attention_layernorm/vars": "model_00001.weights.h5",
|
| 207 |
+
"/layers/qwen3_transformer_decoder_18/_feedforward_gate_dense/vars": "model_00001.weights.h5",
|
| 208 |
+
"/layers/qwen3_transformer_decoder_18/_feedforward_intermediate_dense/vars": "model_00001.weights.h5",
|
| 209 |
+
"/layers/qwen3_transformer_decoder_18/_feedforward_layernorm/vars": "model_00001.weights.h5",
|
| 210 |
+
"/layers/qwen3_transformer_decoder_18/_feedforward_output_dense/vars": "model_00001.weights.h5",
|
| 211 |
+
"/layers/qwen3_transformer_decoder_18/_self_attention_layer/_key_dense/vars": "model_00001.weights.h5",
|
| 212 |
+
"/layers/qwen3_transformer_decoder_18/_self_attention_layer/_key_dense_layer_norm/vars": "model_00001.weights.h5",
|
| 213 |
+
"/layers/qwen3_transformer_decoder_18/_self_attention_layer/_output_dense/vars": "model_00001.weights.h5",
|
| 214 |
+
"/layers/qwen3_transformer_decoder_18/_self_attention_layer/_query_dense/vars": "model_00001.weights.h5",
|
| 215 |
+
"/layers/qwen3_transformer_decoder_18/_self_attention_layer/_query_dense_layer_norm/vars": "model_00001.weights.h5",
|
| 216 |
+
"/layers/qwen3_transformer_decoder_18/_self_attention_layer/_value_dense/vars": "model_00001.weights.h5",
|
| 217 |
+
"/layers/qwen3_transformer_decoder_18/_self_attention_layernorm/vars": "model_00001.weights.h5",
|
| 218 |
+
"/layers/qwen3_transformer_decoder_19/_feedforward_gate_dense/vars": "model_00001.weights.h5",
|
| 219 |
+
"/layers/qwen3_transformer_decoder_19/_feedforward_intermediate_dense/vars": "model_00001.weights.h5",
|
| 220 |
+
"/layers/qwen3_transformer_decoder_19/_feedforward_layernorm/vars": "model_00001.weights.h5",
|
| 221 |
+
"/layers/qwen3_transformer_decoder_19/_feedforward_output_dense/vars": "model_00001.weights.h5",
|
| 222 |
+
"/layers/qwen3_transformer_decoder_19/_self_attention_layer/_key_dense/vars": "model_00001.weights.h5",
|
| 223 |
+
"/layers/qwen3_transformer_decoder_19/_self_attention_layer/_key_dense_layer_norm/vars": "model_00001.weights.h5",
|
| 224 |
+
"/layers/qwen3_transformer_decoder_19/_self_attention_layer/_output_dense/vars": "model_00001.weights.h5",
|
| 225 |
+
"/layers/qwen3_transformer_decoder_19/_self_attention_layer/_query_dense/vars": "model_00001.weights.h5",
|
| 226 |
+
"/layers/qwen3_transformer_decoder_19/_self_attention_layer/_query_dense_layer_norm/vars": "model_00001.weights.h5",
|
| 227 |
+
"/layers/qwen3_transformer_decoder_19/_self_attention_layer/_value_dense/vars": "model_00001.weights.h5",
|
| 228 |
+
"/layers/qwen3_transformer_decoder_19/_self_attention_layernorm/vars": "model_00001.weights.h5",
|
| 229 |
+
"/layers/qwen3_transformer_decoder_20/_feedforward_gate_dense/vars": "model_00001.weights.h5",
|
| 230 |
+
"/layers/qwen3_transformer_decoder_20/_feedforward_intermediate_dense/vars": "model_00001.weights.h5",
|
| 231 |
+
"/layers/qwen3_transformer_decoder_20/_feedforward_layernorm/vars": "model_00001.weights.h5",
|
| 232 |
+
"/layers/qwen3_transformer_decoder_20/_feedforward_output_dense/vars": "model_00001.weights.h5",
|
| 233 |
+
"/layers/qwen3_transformer_decoder_20/_self_attention_layer/_key_dense/vars": "model_00001.weights.h5",
|
| 234 |
+
"/layers/qwen3_transformer_decoder_20/_self_attention_layer/_key_dense_layer_norm/vars": "model_00001.weights.h5",
|
| 235 |
+
"/layers/qwen3_transformer_decoder_20/_self_attention_layer/_output_dense/vars": "model_00001.weights.h5",
|
| 236 |
+
"/layers/qwen3_transformer_decoder_20/_self_attention_layer/_query_dense/vars": "model_00001.weights.h5",
|
| 237 |
+
"/layers/qwen3_transformer_decoder_20/_self_attention_layer/_query_dense_layer_norm/vars": "model_00001.weights.h5",
|
| 238 |
+
"/layers/qwen3_transformer_decoder_20/_self_attention_layer/_value_dense/vars": "model_00001.weights.h5",
|
| 239 |
+
"/layers/qwen3_transformer_decoder_20/_self_attention_layernorm/vars": "model_00001.weights.h5",
|
| 240 |
+
"/layers/qwen3_transformer_decoder_21/_feedforward_gate_dense/vars": "model_00002.weights.h5",
|
| 241 |
+
"/layers/qwen3_transformer_decoder_21/_feedforward_intermediate_dense/vars": "model_00002.weights.h5",
|
| 242 |
+
"/layers/qwen3_transformer_decoder_21/_feedforward_layernorm/vars": "model_00002.weights.h5",
|
| 243 |
+
"/layers/qwen3_transformer_decoder_21/_feedforward_output_dense/vars": "model_00002.weights.h5",
|
| 244 |
+
"/layers/qwen3_transformer_decoder_21/_self_attention_layer/_key_dense/vars": "model_00002.weights.h5",
|
| 245 |
+
"/layers/qwen3_transformer_decoder_21/_self_attention_layer/_key_dense_layer_norm/vars": "model_00002.weights.h5",
|
| 246 |
+
"/layers/qwen3_transformer_decoder_21/_self_attention_layer/_output_dense/vars": "model_00002.weights.h5",
|
| 247 |
+
"/layers/qwen3_transformer_decoder_21/_self_attention_layer/_query_dense/vars": "model_00002.weights.h5",
|
| 248 |
+
"/layers/qwen3_transformer_decoder_21/_self_attention_layer/_query_dense_layer_norm/vars": "model_00002.weights.h5",
|
| 249 |
+
"/layers/qwen3_transformer_decoder_21/_self_attention_layer/_value_dense/vars": "model_00002.weights.h5",
|
| 250 |
+
"/layers/qwen3_transformer_decoder_21/_self_attention_layernorm/vars": "model_00002.weights.h5",
|
| 251 |
+
"/layers/qwen3_transformer_decoder_22/_feedforward_gate_dense/vars": "model_00002.weights.h5",
|
| 252 |
+
"/layers/qwen3_transformer_decoder_22/_feedforward_intermediate_dense/vars": "model_00002.weights.h5",
|
| 253 |
+
"/layers/qwen3_transformer_decoder_22/_feedforward_layernorm/vars": "model_00002.weights.h5",
|
| 254 |
+
"/layers/qwen3_transformer_decoder_22/_feedforward_output_dense/vars": "model_00002.weights.h5",
|
| 255 |
+
"/layers/qwen3_transformer_decoder_22/_self_attention_layer/_key_dense/vars": "model_00002.weights.h5",
|
| 256 |
+
"/layers/qwen3_transformer_decoder_22/_self_attention_layer/_key_dense_layer_norm/vars": "model_00002.weights.h5",
|
| 257 |
+
"/layers/qwen3_transformer_decoder_22/_self_attention_layer/_output_dense/vars": "model_00002.weights.h5",
|
| 258 |
+
"/layers/qwen3_transformer_decoder_22/_self_attention_layer/_query_dense/vars": "model_00002.weights.h5",
|
| 259 |
+
"/layers/qwen3_transformer_decoder_22/_self_attention_layer/_query_dense_layer_norm/vars": "model_00002.weights.h5",
|
| 260 |
+
"/layers/qwen3_transformer_decoder_22/_self_attention_layer/_value_dense/vars": "model_00002.weights.h5",
|
| 261 |
+
"/layers/qwen3_transformer_decoder_22/_self_attention_layernorm/vars": "model_00002.weights.h5",
|
| 262 |
+
"/layers/qwen3_transformer_decoder_23/_feedforward_gate_dense/vars": "model_00002.weights.h5",
|
| 263 |
+
"/layers/qwen3_transformer_decoder_23/_feedforward_intermediate_dense/vars": "model_00002.weights.h5",
|
| 264 |
+
"/layers/qwen3_transformer_decoder_23/_feedforward_layernorm/vars": "model_00002.weights.h5",
|
| 265 |
+
"/layers/qwen3_transformer_decoder_23/_feedforward_output_dense/vars": "model_00002.weights.h5",
|
| 266 |
+
"/layers/qwen3_transformer_decoder_23/_self_attention_layer/_key_dense/vars": "model_00002.weights.h5",
|
| 267 |
+
"/layers/qwen3_transformer_decoder_23/_self_attention_layer/_key_dense_layer_norm/vars": "model_00002.weights.h5",
|
| 268 |
+
"/layers/qwen3_transformer_decoder_23/_self_attention_layer/_output_dense/vars": "model_00002.weights.h5",
|
| 269 |
+
"/layers/qwen3_transformer_decoder_23/_self_attention_layer/_query_dense/vars": "model_00002.weights.h5",
|
| 270 |
+
"/layers/qwen3_transformer_decoder_23/_self_attention_layer/_query_dense_layer_norm/vars": "model_00002.weights.h5",
|
| 271 |
+
"/layers/qwen3_transformer_decoder_23/_self_attention_layer/_value_dense/vars": "model_00002.weights.h5",
|
| 272 |
+
"/layers/qwen3_transformer_decoder_23/_self_attention_layernorm/vars": "model_00002.weights.h5",
|
| 273 |
+
"/layers/qwen3_transformer_decoder_24/_feedforward_gate_dense/vars": "model_00002.weights.h5",
|
| 274 |
+
"/layers/qwen3_transformer_decoder_24/_feedforward_intermediate_dense/vars": "model_00002.weights.h5",
|
| 275 |
+
"/layers/qwen3_transformer_decoder_24/_feedforward_layernorm/vars": "model_00002.weights.h5",
|
| 276 |
+
"/layers/qwen3_transformer_decoder_24/_feedforward_output_dense/vars": "model_00002.weights.h5",
|
| 277 |
+
"/layers/qwen3_transformer_decoder_24/_self_attention_layer/_key_dense/vars": "model_00002.weights.h5",
|
| 278 |
+
"/layers/qwen3_transformer_decoder_24/_self_attention_layer/_key_dense_layer_norm/vars": "model_00002.weights.h5",
|
| 279 |
+
"/layers/qwen3_transformer_decoder_24/_self_attention_layer/_output_dense/vars": "model_00002.weights.h5",
|
| 280 |
+
"/layers/qwen3_transformer_decoder_24/_self_attention_layer/_query_dense/vars": "model_00002.weights.h5",
|
| 281 |
+
"/layers/qwen3_transformer_decoder_24/_self_attention_layer/_query_dense_layer_norm/vars": "model_00002.weights.h5",
|
| 282 |
+
"/layers/qwen3_transformer_decoder_24/_self_attention_layer/_value_dense/vars": "model_00002.weights.h5",
|
| 283 |
+
"/layers/qwen3_transformer_decoder_24/_self_attention_layernorm/vars": "model_00002.weights.h5",
|
| 284 |
+
"/layers/qwen3_transformer_decoder_25/_feedforward_gate_dense/vars": "model_00002.weights.h5",
|
| 285 |
+
"/layers/qwen3_transformer_decoder_25/_feedforward_intermediate_dense/vars": "model_00002.weights.h5",
|
| 286 |
+
"/layers/qwen3_transformer_decoder_25/_feedforward_layernorm/vars": "model_00002.weights.h5",
|
| 287 |
+
"/layers/qwen3_transformer_decoder_25/_feedforward_output_dense/vars": "model_00002.weights.h5",
|
| 288 |
+
"/layers/qwen3_transformer_decoder_25/_self_attention_layer/_key_dense/vars": "model_00002.weights.h5",
|
| 289 |
+
"/layers/qwen3_transformer_decoder_25/_self_attention_layer/_key_dense_layer_norm/vars": "model_00002.weights.h5",
|
| 290 |
+
"/layers/qwen3_transformer_decoder_25/_self_attention_layer/_output_dense/vars": "model_00002.weights.h5",
|
| 291 |
+
"/layers/qwen3_transformer_decoder_25/_self_attention_layer/_query_dense/vars": "model_00002.weights.h5",
|
| 292 |
+
"/layers/qwen3_transformer_decoder_25/_self_attention_layer/_query_dense_layer_norm/vars": "model_00002.weights.h5",
|
| 293 |
+
"/layers/qwen3_transformer_decoder_25/_self_attention_layer/_value_dense/vars": "model_00002.weights.h5",
|
| 294 |
+
"/layers/qwen3_transformer_decoder_25/_self_attention_layernorm/vars": "model_00002.weights.h5",
|
| 295 |
+
"/layers/qwen3_transformer_decoder_26/_feedforward_gate_dense/vars": "model_00002.weights.h5",
|
| 296 |
+
"/layers/qwen3_transformer_decoder_26/_feedforward_intermediate_dense/vars": "model_00002.weights.h5",
|
| 297 |
+
"/layers/qwen3_transformer_decoder_26/_feedforward_layernorm/vars": "model_00002.weights.h5",
|
| 298 |
+
"/layers/qwen3_transformer_decoder_26/_feedforward_output_dense/vars": "model_00002.weights.h5",
|
| 299 |
+
"/layers/qwen3_transformer_decoder_26/_self_attention_layer/_key_dense/vars": "model_00002.weights.h5",
|
| 300 |
+
"/layers/qwen3_transformer_decoder_26/_self_attention_layer/_key_dense_layer_norm/vars": "model_00002.weights.h5",
|
| 301 |
+
"/layers/qwen3_transformer_decoder_26/_self_attention_layer/_output_dense/vars": "model_00002.weights.h5",
|
| 302 |
+
"/layers/qwen3_transformer_decoder_26/_self_attention_layer/_query_dense/vars": "model_00002.weights.h5",
|
| 303 |
+
"/layers/qwen3_transformer_decoder_26/_self_attention_layer/_query_dense_layer_norm/vars": "model_00002.weights.h5",
|
| 304 |
+
"/layers/qwen3_transformer_decoder_26/_self_attention_layer/_value_dense/vars": "model_00002.weights.h5",
|
| 305 |
+
"/layers/qwen3_transformer_decoder_26/_self_attention_layernorm/vars": "model_00002.weights.h5",
|
| 306 |
+
"/layers/qwen3_transformer_decoder_27/_feedforward_gate_dense/vars": "model_00002.weights.h5",
|
| 307 |
+
"/layers/qwen3_transformer_decoder_27/_feedforward_intermediate_dense/vars": "model_00002.weights.h5",
|
| 308 |
+
"/layers/qwen3_transformer_decoder_27/_feedforward_layernorm/vars": "model_00002.weights.h5",
|
| 309 |
+
"/layers/qwen3_transformer_decoder_27/_feedforward_output_dense/vars": "model_00002.weights.h5",
|
| 310 |
+
"/layers/qwen3_transformer_decoder_27/_self_attention_layer/_key_dense/vars": "model_00002.weights.h5",
|
| 311 |
+
"/layers/qwen3_transformer_decoder_27/_self_attention_layer/_key_dense_layer_norm/vars": "model_00002.weights.h5",
|
| 312 |
+
"/layers/qwen3_transformer_decoder_27/_self_attention_layer/_output_dense/vars": "model_00002.weights.h5",
|
| 313 |
+
"/layers/qwen3_transformer_decoder_27/_self_attention_layer/_query_dense/vars": "model_00002.weights.h5",
|
| 314 |
+
"/layers/qwen3_transformer_decoder_27/_self_attention_layer/_query_dense_layer_norm/vars": "model_00002.weights.h5",
|
| 315 |
+
"/layers/qwen3_transformer_decoder_27/_self_attention_layer/_value_dense/vars": "model_00002.weights.h5",
|
| 316 |
+
"/layers/qwen3_transformer_decoder_27/_self_attention_layernorm/vars": "model_00002.weights.h5",
|
| 317 |
+
"/layers/qwen3_transformer_decoder_28/_feedforward_gate_dense/vars": "model_00002.weights.h5",
|
| 318 |
+
"/layers/qwen3_transformer_decoder_28/_feedforward_intermediate_dense/vars": "model_00002.weights.h5",
|
| 319 |
+
"/layers/qwen3_transformer_decoder_28/_feedforward_layernorm/vars": "model_00002.weights.h5",
|
| 320 |
+
"/layers/qwen3_transformer_decoder_28/_feedforward_output_dense/vars": "model_00002.weights.h5",
|
| 321 |
+
"/layers/qwen3_transformer_decoder_28/_self_attention_layer/_key_dense/vars": "model_00002.weights.h5",
|
| 322 |
+
"/layers/qwen3_transformer_decoder_28/_self_attention_layer/_key_dense_layer_norm/vars": "model_00002.weights.h5",
|
| 323 |
+
"/layers/qwen3_transformer_decoder_28/_self_attention_layer/_output_dense/vars": "model_00002.weights.h5",
|
| 324 |
+
"/layers/qwen3_transformer_decoder_28/_self_attention_layer/_query_dense/vars": "model_00002.weights.h5",
|
| 325 |
+
"/layers/qwen3_transformer_decoder_28/_self_attention_layer/_query_dense_layer_norm/vars": "model_00002.weights.h5",
|
| 326 |
+
"/layers/qwen3_transformer_decoder_28/_self_attention_layer/_value_dense/vars": "model_00002.weights.h5",
|
| 327 |
+
"/layers/qwen3_transformer_decoder_28/_self_attention_layernorm/vars": "model_00002.weights.h5",
|
| 328 |
+
"/layers/qwen3_transformer_decoder_29/_feedforward_gate_dense/vars": "model_00002.weights.h5",
|
| 329 |
+
"/layers/qwen3_transformer_decoder_29/_feedforward_intermediate_dense/vars": "model_00002.weights.h5",
|
| 330 |
+
"/layers/qwen3_transformer_decoder_29/_feedforward_layernorm/vars": "model_00002.weights.h5",
|
| 331 |
+
"/layers/qwen3_transformer_decoder_29/_feedforward_output_dense/vars": "model_00002.weights.h5",
|
| 332 |
+
"/layers/qwen3_transformer_decoder_29/_self_attention_layer/_key_dense/vars": "model_00002.weights.h5",
|
| 333 |
+
"/layers/qwen3_transformer_decoder_29/_self_attention_layer/_key_dense_layer_norm/vars": "model_00002.weights.h5",
|
| 334 |
+
"/layers/qwen3_transformer_decoder_29/_self_attention_layer/_output_dense/vars": "model_00002.weights.h5",
|
| 335 |
+
"/layers/qwen3_transformer_decoder_29/_self_attention_layer/_query_dense/vars": "model_00002.weights.h5",
|
| 336 |
+
"/layers/qwen3_transformer_decoder_29/_self_attention_layer/_query_dense_layer_norm/vars": "model_00002.weights.h5",
|
| 337 |
+
"/layers/qwen3_transformer_decoder_29/_self_attention_layer/_value_dense/vars": "model_00002.weights.h5",
|
| 338 |
+
"/layers/qwen3_transformer_decoder_29/_self_attention_layernorm/vars": "model_00002.weights.h5",
|
| 339 |
+
"/layers/qwen3_transformer_decoder_30/_feedforward_gate_dense/vars": "model_00002.weights.h5",
|
| 340 |
+
"/layers/qwen3_transformer_decoder_30/_feedforward_intermediate_dense/vars": "model_00002.weights.h5",
|
| 341 |
+
"/layers/qwen3_transformer_decoder_30/_feedforward_layernorm/vars": "model_00002.weights.h5",
|
| 342 |
+
"/layers/qwen3_transformer_decoder_30/_feedforward_output_dense/vars": "model_00002.weights.h5",
|
| 343 |
+
"/layers/qwen3_transformer_decoder_30/_self_attention_layer/_key_dense/vars": "model_00002.weights.h5",
|
| 344 |
+
"/layers/qwen3_transformer_decoder_30/_self_attention_layer/_key_dense_layer_norm/vars": "model_00002.weights.h5",
|
| 345 |
+
"/layers/qwen3_transformer_decoder_30/_self_attention_layer/_output_dense/vars": "model_00002.weights.h5",
|
| 346 |
+
"/layers/qwen3_transformer_decoder_30/_self_attention_layer/_query_dense/vars": "model_00002.weights.h5",
|
| 347 |
+
"/layers/qwen3_transformer_decoder_30/_self_attention_layer/_query_dense_layer_norm/vars": "model_00002.weights.h5",
|
| 348 |
+
"/layers/qwen3_transformer_decoder_30/_self_attention_layer/_value_dense/vars": "model_00002.weights.h5",
|
| 349 |
+
"/layers/qwen3_transformer_decoder_30/_self_attention_layernorm/vars": "model_00002.weights.h5",
|
| 350 |
+
"/layers/qwen3_transformer_decoder_31/_feedforward_gate_dense/vars": "model_00002.weights.h5",
|
| 351 |
+
"/layers/qwen3_transformer_decoder_31/_feedforward_intermediate_dense/vars": "model_00002.weights.h5",
|
| 352 |
+
"/layers/qwen3_transformer_decoder_31/_feedforward_layernorm/vars": "model_00002.weights.h5",
|
| 353 |
+
"/layers/qwen3_transformer_decoder_31/_feedforward_output_dense/vars": "model_00002.weights.h5",
|
| 354 |
+
"/layers/qwen3_transformer_decoder_31/_self_attention_layer/_key_dense/vars": "model_00002.weights.h5",
|
| 355 |
+
"/layers/qwen3_transformer_decoder_31/_self_attention_layer/_key_dense_layer_norm/vars": "model_00002.weights.h5",
|
| 356 |
+
"/layers/qwen3_transformer_decoder_31/_self_attention_layer/_output_dense/vars": "model_00002.weights.h5",
|
| 357 |
+
"/layers/qwen3_transformer_decoder_31/_self_attention_layer/_query_dense/vars": "model_00002.weights.h5",
|
| 358 |
+
"/layers/qwen3_transformer_decoder_31/_self_attention_layer/_query_dense_layer_norm/vars": "model_00002.weights.h5",
|
| 359 |
+
"/layers/qwen3_transformer_decoder_31/_self_attention_layer/_value_dense/vars": "model_00002.weights.h5",
|
| 360 |
+
"/layers/qwen3_transformer_decoder_31/_self_attention_layernorm/vars": "model_00002.weights.h5",
|
| 361 |
+
"/layers/qwen3_transformer_decoder_32/_feedforward_gate_dense/vars": "model_00002.weights.h5",
|
| 362 |
+
"/layers/qwen3_transformer_decoder_32/_feedforward_intermediate_dense/vars": "model_00002.weights.h5",
|
| 363 |
+
"/layers/qwen3_transformer_decoder_32/_feedforward_layernorm/vars": "model_00002.weights.h5",
|
| 364 |
+
"/layers/qwen3_transformer_decoder_32/_feedforward_output_dense/vars": "model_00002.weights.h5",
|
| 365 |
+
"/layers/qwen3_transformer_decoder_32/_self_attention_layer/_key_dense/vars": "model_00002.weights.h5",
|
| 366 |
+
"/layers/qwen3_transformer_decoder_32/_self_attention_layer/_key_dense_layer_norm/vars": "model_00002.weights.h5",
|
| 367 |
+
"/layers/qwen3_transformer_decoder_32/_self_attention_layer/_output_dense/vars": "model_00002.weights.h5",
|
| 368 |
+
"/layers/qwen3_transformer_decoder_32/_self_attention_layer/_query_dense/vars": "model_00002.weights.h5",
|
| 369 |
+
"/layers/qwen3_transformer_decoder_32/_self_attention_layer/_query_dense_layer_norm/vars": "model_00002.weights.h5",
|
| 370 |
+
"/layers/qwen3_transformer_decoder_32/_self_attention_layer/_value_dense/vars": "model_00002.weights.h5",
|
| 371 |
+
"/layers/qwen3_transformer_decoder_32/_self_attention_layernorm/vars": "model_00002.weights.h5",
|
| 372 |
+
"/layers/qwen3_transformer_decoder_33/_feedforward_gate_dense/vars": "model_00002.weights.h5",
|
| 373 |
+
"/layers/qwen3_transformer_decoder_33/_feedforward_intermediate_dense/vars": "model_00002.weights.h5",
|
| 374 |
+
"/layers/qwen3_transformer_decoder_33/_feedforward_layernorm/vars": "model_00002.weights.h5",
|
| 375 |
+
"/layers/qwen3_transformer_decoder_33/_feedforward_output_dense/vars": "model_00002.weights.h5",
|
| 376 |
+
"/layers/qwen3_transformer_decoder_33/_self_attention_layer/_key_dense/vars": "model_00002.weights.h5",
|
| 377 |
+
"/layers/qwen3_transformer_decoder_33/_self_attention_layer/_key_dense_layer_norm/vars": "model_00002.weights.h5",
|
| 378 |
+
"/layers/qwen3_transformer_decoder_33/_self_attention_layer/_output_dense/vars": "model_00002.weights.h5",
|
| 379 |
+
"/layers/qwen3_transformer_decoder_33/_self_attention_layer/_query_dense/vars": "model_00002.weights.h5",
|
| 380 |
+
"/layers/qwen3_transformer_decoder_33/_self_attention_layer/_query_dense_layer_norm/vars": "model_00002.weights.h5",
|
| 381 |
+
"/layers/qwen3_transformer_decoder_33/_self_attention_layer/_value_dense/vars": "model_00002.weights.h5",
|
| 382 |
+
"/layers/qwen3_transformer_decoder_33/_self_attention_layernorm/vars": "model_00002.weights.h5",
|
| 383 |
+
"/layers/qwen3_transformer_decoder_34/_feedforward_gate_dense/vars": "model_00002.weights.h5",
|
| 384 |
+
"/layers/qwen3_transformer_decoder_34/_feedforward_intermediate_dense/vars": "model_00002.weights.h5",
|
| 385 |
+
"/layers/qwen3_transformer_decoder_34/_feedforward_layernorm/vars": "model_00002.weights.h5",
|
| 386 |
+
"/layers/qwen3_transformer_decoder_34/_feedforward_output_dense/vars": "model_00002.weights.h5",
|
| 387 |
+
"/layers/qwen3_transformer_decoder_34/_self_attention_layer/_key_dense/vars": "model_00002.weights.h5",
|
| 388 |
+
"/layers/qwen3_transformer_decoder_34/_self_attention_layer/_key_dense_layer_norm/vars": "model_00002.weights.h5",
|
| 389 |
+
"/layers/qwen3_transformer_decoder_34/_self_attention_layer/_output_dense/vars": "model_00002.weights.h5",
|
| 390 |
+
"/layers/qwen3_transformer_decoder_34/_self_attention_layer/_query_dense/vars": "model_00003.weights.h5",
|
| 391 |
+
"/layers/qwen3_transformer_decoder_34/_self_attention_layer/_query_dense_layer_norm/vars": "model_00003.weights.h5",
|
| 392 |
+
"/layers/qwen3_transformer_decoder_34/_self_attention_layer/_value_dense/vars": "model_00003.weights.h5",
|
| 393 |
+
"/layers/qwen3_transformer_decoder_34/_self_attention_layernorm/vars": "model_00003.weights.h5",
|
| 394 |
+
"/layers/qwen3_transformer_decoder_35/_feedforward_gate_dense/vars": "model_00003.weights.h5",
|
| 395 |
+
"/layers/qwen3_transformer_decoder_35/_feedforward_intermediate_dense/vars": "model_00003.weights.h5",
|
| 396 |
+
"/layers/qwen3_transformer_decoder_35/_feedforward_layernorm/vars": "model_00003.weights.h5",
|
| 397 |
+
"/layers/qwen3_transformer_decoder_35/_feedforward_output_dense/vars": "model_00003.weights.h5",
|
| 398 |
+
"/layers/qwen3_transformer_decoder_35/_self_attention_layer/_key_dense/vars": "model_00003.weights.h5",
|
| 399 |
+
"/layers/qwen3_transformer_decoder_35/_self_attention_layer/_key_dense_layer_norm/vars": "model_00003.weights.h5",
|
| 400 |
+
"/layers/qwen3_transformer_decoder_35/_self_attention_layer/_output_dense/vars": "model_00003.weights.h5",
|
| 401 |
+
"/layers/qwen3_transformer_decoder_35/_self_attention_layer/_query_dense/vars": "model_00003.weights.h5",
|
| 402 |
+
"/layers/qwen3_transformer_decoder_35/_self_attention_layer/_query_dense_layer_norm/vars": "model_00003.weights.h5",
|
| 403 |
+
"/layers/qwen3_transformer_decoder_35/_self_attention_layer/_value_dense/vars": "model_00003.weights.h5",
|
| 404 |
+
"/layers/qwen3_transformer_decoder_35/_self_attention_layernorm/vars": "model_00003.weights.h5",
|
| 405 |
+
"/layers/qwen3_layer_norm/vars": "model_00003.weights.h5"
|
| 406 |
+
}
|
| 407 |
+
}
|
model_00000.weights.h5
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:49f3f5fcd7436d4c19448edd80a7496dce7d94c7bf1f19758c75d4e396bc29f3
|
| 3 |
+
size 10582761328
|
model_00001.weights.h5
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6217a6b719ec6c022f48417a07b88eae7e749da3f636ac32077965e5852a2612
|
| 3 |
+
size 10604230600
|
model_00002.weights.h5
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:743b85a5f2a7489c6938e26d392f449d465359d479d6440c9e5abd250a199ba4
|
| 3 |
+
size 10721643272
|
model_00003.weights.h5
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7a1bed266705905c8544a105756c19fa04ff96495004c00aa432682b1a2a54c8
|
| 3 |
+
size 855766696
|
preprocessor.json
ADDED
|
@@ -0,0 +1,69 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"module": "keras_hub.src.models.qwen3.qwen3_causal_lm_preprocessor",
|
| 3 |
+
"class_name": "Qwen3CausalLMPreprocessor",
|
| 4 |
+
"config": {
|
| 5 |
+
"name": "qwen3_causal_lm_preprocessor_2",
|
| 6 |
+
"trainable": true,
|
| 7 |
+
"dtype": {
|
| 8 |
+
"module": "keras",
|
| 9 |
+
"class_name": "DTypePolicy",
|
| 10 |
+
"config": {
|
| 11 |
+
"name": "float32"
|
| 12 |
+
},
|
| 13 |
+
"registered_name": null
|
| 14 |
+
},
|
| 15 |
+
"tokenizer": {
|
| 16 |
+
"module": "keras_hub.src.models.qwen3.qwen3_tokenizer",
|
| 17 |
+
"class_name": "Qwen3Tokenizer",
|
| 18 |
+
"config": {
|
| 19 |
+
"name": "qwen3_tokenizer",
|
| 20 |
+
"trainable": true,
|
| 21 |
+
"dtype": {
|
| 22 |
+
"module": "keras",
|
| 23 |
+
"class_name": "DTypePolicy",
|
| 24 |
+
"config": {
|
| 25 |
+
"name": "int32"
|
| 26 |
+
},
|
| 27 |
+
"registered_name": null
|
| 28 |
+
},
|
| 29 |
+
"config_file": "tokenizer.json",
|
| 30 |
+
"sequence_length": null,
|
| 31 |
+
"add_prefix_space": false,
|
| 32 |
+
"unsplittable_tokens": [
|
| 33 |
+
"</think>",
|
| 34 |
+
"<|endoftext|>",
|
| 35 |
+
"<|vision_end|>",
|
| 36 |
+
"<|vision_start|>",
|
| 37 |
+
"<|quad_end|>",
|
| 38 |
+
"<|im_end|>",
|
| 39 |
+
"<|image_pad|>",
|
| 40 |
+
"<|box_end|>",
|
| 41 |
+
"</tool_response>",
|
| 42 |
+
"<|box_start|>",
|
| 43 |
+
"</tool_call>",
|
| 44 |
+
"<|fim_suffix|>",
|
| 45 |
+
"<think>",
|
| 46 |
+
"<|fim_prefix|>",
|
| 47 |
+
"<|quad_start|>",
|
| 48 |
+
"<|repo_name|>",
|
| 49 |
+
"<|fim_middle|>",
|
| 50 |
+
"<|object_ref_end|>",
|
| 51 |
+
"<|object_ref_start|>",
|
| 52 |
+
"<tool_call>",
|
| 53 |
+
"<|video_pad|>",
|
| 54 |
+
"<tool_response>",
|
| 55 |
+
"<|vision_pad|>",
|
| 56 |
+
"<|im_start|>",
|
| 57 |
+
"<|file_sep|>",
|
| 58 |
+
"<|fim_pad|>"
|
| 59 |
+
]
|
| 60 |
+
},
|
| 61 |
+
"registered_name": "keras_hub>Qwen3Tokenizer"
|
| 62 |
+
},
|
| 63 |
+
"config_file": "preprocessor.json",
|
| 64 |
+
"sequence_length": 1024,
|
| 65 |
+
"add_start_token": true,
|
| 66 |
+
"add_end_token": true
|
| 67 |
+
},
|
| 68 |
+
"registered_name": "keras_hub>Qwen3CausalLMPreprocessor"
|
| 69 |
+
}
|
task.json
ADDED
|
@@ -0,0 +1,99 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"module": "keras_hub.src.models.qwen3.qwen3_causal_lm",
|
| 3 |
+
"class_name": "Qwen3CausalLM",
|
| 4 |
+
"config": {
|
| 5 |
+
"backbone": {
|
| 6 |
+
"module": "keras_hub.src.models.qwen3.qwen3_backbone",
|
| 7 |
+
"class_name": "Qwen3Backbone",
|
| 8 |
+
"config": {
|
| 9 |
+
"name": "qwen3_backbone",
|
| 10 |
+
"trainable": true,
|
| 11 |
+
"vocabulary_size": 151936,
|
| 12 |
+
"num_layers": 36,
|
| 13 |
+
"num_query_heads": 32,
|
| 14 |
+
"hidden_dim": 4096,
|
| 15 |
+
"head_dim": 128,
|
| 16 |
+
"intermediate_dim": 12288,
|
| 17 |
+
"rope_max_wavelength": 1000000,
|
| 18 |
+
"rope_scaling_factor": 1.0,
|
| 19 |
+
"num_key_value_heads": 8,
|
| 20 |
+
"layer_norm_epsilon": 1e-06,
|
| 21 |
+
"dropout": 0.0,
|
| 22 |
+
"tie_word_embeddings": false,
|
| 23 |
+
"sliding_window_size": null
|
| 24 |
+
},
|
| 25 |
+
"registered_name": "keras_hub>Qwen3Backbone"
|
| 26 |
+
},
|
| 27 |
+
"preprocessor": {
|
| 28 |
+
"module": "keras_hub.src.models.qwen3.qwen3_causal_lm_preprocessor",
|
| 29 |
+
"class_name": "Qwen3CausalLMPreprocessor",
|
| 30 |
+
"config": {
|
| 31 |
+
"name": "qwen3_causal_lm_preprocessor_2",
|
| 32 |
+
"trainable": true,
|
| 33 |
+
"dtype": {
|
| 34 |
+
"module": "keras",
|
| 35 |
+
"class_name": "DTypePolicy",
|
| 36 |
+
"config": {
|
| 37 |
+
"name": "float32"
|
| 38 |
+
},
|
| 39 |
+
"registered_name": null
|
| 40 |
+
},
|
| 41 |
+
"tokenizer": {
|
| 42 |
+
"module": "keras_hub.src.models.qwen3.qwen3_tokenizer",
|
| 43 |
+
"class_name": "Qwen3Tokenizer",
|
| 44 |
+
"config": {
|
| 45 |
+
"name": "qwen3_tokenizer",
|
| 46 |
+
"trainable": true,
|
| 47 |
+
"dtype": {
|
| 48 |
+
"module": "keras",
|
| 49 |
+
"class_name": "DTypePolicy",
|
| 50 |
+
"config": {
|
| 51 |
+
"name": "int32"
|
| 52 |
+
},
|
| 53 |
+
"registered_name": null
|
| 54 |
+
},
|
| 55 |
+
"config_file": "tokenizer.json",
|
| 56 |
+
"sequence_length": null,
|
| 57 |
+
"add_prefix_space": false,
|
| 58 |
+
"unsplittable_tokens": [
|
| 59 |
+
"</think>",
|
| 60 |
+
"<|endoftext|>",
|
| 61 |
+
"<|vision_end|>",
|
| 62 |
+
"<|vision_start|>",
|
| 63 |
+
"<|quad_end|>",
|
| 64 |
+
"<|im_end|>",
|
| 65 |
+
"<|image_pad|>",
|
| 66 |
+
"<|box_end|>",
|
| 67 |
+
"</tool_response>",
|
| 68 |
+
"<|box_start|>",
|
| 69 |
+
"</tool_call>",
|
| 70 |
+
"<|fim_suffix|>",
|
| 71 |
+
"<think>",
|
| 72 |
+
"<|fim_prefix|>",
|
| 73 |
+
"<|quad_start|>",
|
| 74 |
+
"<|repo_name|>",
|
| 75 |
+
"<|fim_middle|>",
|
| 76 |
+
"<|object_ref_end|>",
|
| 77 |
+
"<|object_ref_start|>",
|
| 78 |
+
"<tool_call>",
|
| 79 |
+
"<|video_pad|>",
|
| 80 |
+
"<tool_response>",
|
| 81 |
+
"<|vision_pad|>",
|
| 82 |
+
"<|im_start|>",
|
| 83 |
+
"<|file_sep|>",
|
| 84 |
+
"<|fim_pad|>"
|
| 85 |
+
]
|
| 86 |
+
},
|
| 87 |
+
"registered_name": "keras_hub>Qwen3Tokenizer"
|
| 88 |
+
},
|
| 89 |
+
"config_file": "preprocessor.json",
|
| 90 |
+
"sequence_length": 1024,
|
| 91 |
+
"add_start_token": true,
|
| 92 |
+
"add_end_token": true
|
| 93 |
+
},
|
| 94 |
+
"registered_name": "keras_hub>Qwen3CausalLMPreprocessor"
|
| 95 |
+
},
|
| 96 |
+
"name": "qwen3_causal_lm"
|
| 97 |
+
},
|
| 98 |
+
"registered_name": "keras_hub>Qwen3CausalLM"
|
| 99 |
+
}
|
tokenizer.json
ADDED
|
@@ -0,0 +1,48 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"module": "keras_hub.src.models.qwen3.qwen3_tokenizer",
|
| 3 |
+
"class_name": "Qwen3Tokenizer",
|
| 4 |
+
"config": {
|
| 5 |
+
"name": "qwen3_tokenizer",
|
| 6 |
+
"trainable": true,
|
| 7 |
+
"dtype": {
|
| 8 |
+
"module": "keras",
|
| 9 |
+
"class_name": "DTypePolicy",
|
| 10 |
+
"config": {
|
| 11 |
+
"name": "int32"
|
| 12 |
+
},
|
| 13 |
+
"registered_name": null
|
| 14 |
+
},
|
| 15 |
+
"config_file": "tokenizer.json",
|
| 16 |
+
"sequence_length": null,
|
| 17 |
+
"add_prefix_space": false,
|
| 18 |
+
"unsplittable_tokens": [
|
| 19 |
+
"</think>",
|
| 20 |
+
"<|endoftext|>",
|
| 21 |
+
"<|vision_end|>",
|
| 22 |
+
"<|vision_start|>",
|
| 23 |
+
"<|quad_end|>",
|
| 24 |
+
"<|im_end|>",
|
| 25 |
+
"<|image_pad|>",
|
| 26 |
+
"<|box_end|>",
|
| 27 |
+
"</tool_response>",
|
| 28 |
+
"<|box_start|>",
|
| 29 |
+
"</tool_call>",
|
| 30 |
+
"<|fim_suffix|>",
|
| 31 |
+
"<think>",
|
| 32 |
+
"<|fim_prefix|>",
|
| 33 |
+
"<|quad_start|>",
|
| 34 |
+
"<|repo_name|>",
|
| 35 |
+
"<|fim_middle|>",
|
| 36 |
+
"<|object_ref_end|>",
|
| 37 |
+
"<|object_ref_start|>",
|
| 38 |
+
"<tool_call>",
|
| 39 |
+
"<|video_pad|>",
|
| 40 |
+
"<tool_response>",
|
| 41 |
+
"<|vision_pad|>",
|
| 42 |
+
"<|im_start|>",
|
| 43 |
+
"<|file_sep|>",
|
| 44 |
+
"<|fim_pad|>"
|
| 45 |
+
]
|
| 46 |
+
},
|
| 47 |
+
"registered_name": "keras_hub>Qwen3Tokenizer"
|
| 48 |
+
}
|