Update tiny models for Qwen4ExpTextModel

config.json

@@ -0,0 +1,71 @@
+{
+  "architectures": [
+    "Qwen4ExpTextModel"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 1,
+  "dtype": "float32",
+  "eos_token_id": 2,
+  "hc_count": 2,
+  "hc_lowrank": 8,
+  "head_dim": 16,
+  "heads_per_ngram": 2,
+  "hidden_act": "silu",
+  "hidden_size": 32,
+  "indexer_budget": 4,
+  "indexer_compress_ratio": 2,
+  "indexer_head_dim": 8,
+  "indexer_kv_heads": 1,
+  "indexer_n_heads": 2,
+  "initializer_range": 0.02,
+  "layer_types": [
+    "linear_attention",
+    "indexed_attention"
+  ],
+  "linear_conv_kernel_dim": 2,
+  "linear_key_head_dim": 16,
+  "linear_num_key_heads": 4,
+  "linear_num_value_heads": 8,
+  "linear_value_head_dim": 16,
+  "make_ngram_vocab_size_divisible_by": 8,
+  "max_position_embeddings": 512,
+  "model_type": "qwen4_exp_text",
+  "moe_intermediate_size": 8,
+  "ngram_size": 3,
+  "ngram_vocab_size_base": 31,
+  "norm_topk_prob": true,
+  "num_attention_heads": 2,
+  "num_experts": 4,
+  "num_experts_per_tok": 2,
+  "num_hidden_layers": 2,
+  "num_key_value_heads": 2,
+  "number_of_conv_states": 3,
+  "output_gate_type": null,
+  "output_router_logits": false,
+  "pad_token_id": 0,
+  "ple_conv_kernel_size": 2,
+  "ple_embed_dim": 16,
+  "ple_layer_ids": [
+    1
+  ],
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "mrope_section": [
+      1,
+      1,
+      1
+    ],
+    "partial_rotary_factor": 0.25,
+    "rope_theta": 10000.0,
+    "rope_type": "default"
+  },
+  "router_aux_loss_coef": 0.001,
+  "seed": 1234,
+  "shared_expert_intermediate_size": 8,
+  "split_ngram_parts": 4,
+  "tie_word_embeddings": false,
+  "transformers_version": "5.19.0.dev0",
+  "use_cache": true,
+  "vocab_size": 99
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e97cc478d8f1f3b3e84778203486e9e0e714a3d2d6dfce15b6df1978de2d0c48
+size 180888