Update tiny models for DeepseekV4ForCausalLM

config.json

@@ -0,0 +1,77 @@
+{
+  "architectures": [
+    "DeepseekV4ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 0,
+  "compress_rates": {
+    "compressed_sparse_attention": 4,
+    "heavily_compressed_attention": 128
+  },
+  "compress_rope_theta": 160000.0,
+  "dtype": "float32",
+  "eos_token_id": 1,
+  "hc_eps": 1e-06,
+  "hc_mult": 2,
+  "hc_sinkhorn_iters": 3,
+  "head_dim": 32,
+  "hidden_act": "gelu",
+  "hidden_size": 64,
+  "index_head_dim": 16,
+  "index_n_heads": 2,
+  "index_topk": 2,
+  "initializer_range": 0.02,
+  "is_decoder": true,
+  "layer_types": [
+    "heavily_compressed_attention",
+    "compressed_sparse_attention"
+  ],
+  "max_position_embeddings": 200,
+  "mlp_bias": false,
+  "mlp_layer_types": [
+    "moe",
+    "moe"
+  ],
+  "model_type": "deepseek_v4",
+  "moe_intermediate_size": 64,
+  "n_routed_experts": 4,
+  "n_shared_experts": 1,
+  "norm_topk_prob": true,
+  "num_attention_heads": 4,
+  "num_experts_per_tok": 2,
+  "num_hidden_layers": 2,
+  "num_key_value_heads": 1,
+  "num_nextn_predict_layers": 0,
+  "o_groups": 2,
+  "o_lora_rank": 16,
+  "output_router_logits": false,
+  "pad_token_id": 1,
+  "partial_rotary_factor": 0.25,
+  "q_lora_rank": 32,
+  "qk_rope_head_dim": 8,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "compress": {
+      "partial_rotary_factor": 0.25,
+      "rope_theta": 160000.0,
+      "rope_type": "default"
+    },
+    "main": {
+      "partial_rotary_factor": 0.25,
+      "rope_theta": 10000.0,
+      "rope_type": "default"
+    }
+  },
+  "rope_theta": 10000.0,
+  "routed_scaling_factor": 1.5,
+  "router_aux_loss_coef": 0.001,
+  "router_jitter_noise": 0.0,
+  "scoring_func": "sqrtsoftplus",
+  "sliding_window": 8,
+  "swiglu_limit": 10.0,
+  "tie_word_embeddings": false,
+  "transformers_version": "5.16.0.dev0",
+  "use_cache": true,
+  "vocab_size": 129280
+}

generation_config.json

@@ -0,0 +1,10 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 0,
+  "eos_token_id": 1,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 1,
+  "transformers_version": "5.16.0.dev0",
+  "use_cache": true
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8924be97b4f4f152420fbc5100f343c0e462378155acbcfc8c84e8c014b83bae
+size 66899964

tokenizer_config.json

@@ -0,0 +1,14 @@
+{
+  "backend": "tokenizers",
+  "bos_token": "<｜begin▁of▁sentence｜>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<｜end▁of▁sentence｜>",
+  "is_local": true,
+  "legacy": true,
+  "local_files_only": false,
+  "model_max_length": 200,
+  "pad_token": "<｜end▁of▁sentence｜>",
+  "sp_model_kwargs": {},
+  "tokenizer_class": "TokenizersBackend",
+  "unk_token": null
+}