diff --git a/.gitattributes b/.gitattributes
index a6344aac8c09253b3b630fb776ae94478aa0275b..33c297035c3443c7e913d3bf2b3e2139e22209b6 100644
--- a/.gitattributes
+++ b/.gitattributes
@@ -33,3 +33,23 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+Keen-qwen4b-20b/checkpoint-4453/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+Keen-qwen4b-20b/checkpoint-6701/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+Keen-qwen4b-20b/checkpoint-17953/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+Keen-qwen4b-20b/checkpoint-15698/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+Keen-qwen4b-20b/checkpoint-2253/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+Keen-qwen4b-20b/checkpoint-8956/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+Keen-qwen4b-20b/checkpoint-11202/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+Keen-qwen4b-20b/checkpoint-13449/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+Keen-qwen4b-20b/checkpoint-20218/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+Keen-qwen4b-20b/checkpoint-22478/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+Vanilla-qwen4b-20b/checkpoint-9543/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+Vanilla-qwen4b-20b/checkpoint-1910/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+Vanilla-qwen4b-20b/checkpoint-7635/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+Vanilla-qwen4b-20b/checkpoint-11452/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+Vanilla-qwen4b-20b/checkpoint-19106/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+Vanilla-qwen4b-20b/checkpoint-3818/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+Vanilla-qwen4b-20b/checkpoint-17188/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+Vanilla-qwen4b-20b/checkpoint-5726/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+Vanilla-qwen4b-20b/checkpoint-13362/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+Vanilla-qwen4b-20b/checkpoint-15273/tokenizer.json filter=lfs diff=lfs merge=lfs -text
diff --git a/Keen-qwen4b-20b/checkpoint-11202/model.safetensors b/Keen-qwen4b-20b/checkpoint-11202/model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..770910b69801907d50b23962b0ac4a2bf7c01f69
--- /dev/null
+++ b/Keen-qwen4b-20b/checkpoint-11202/model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9e5c178a11a7f4d1f2c4486a1e21a47acfba0998e2484122c6dbe2e9f61f9101
+size 8411554440
diff --git a/Keen-qwen4b-20b/checkpoint-11202/tokenizer.json b/Keen-qwen4b-20b/checkpoint-11202/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..ec1bde77c388a3d5523769ad975c655ae8ed5eda
--- /dev/null
+++ b/Keen-qwen4b-20b/checkpoint-11202/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4f8a9ce416e1039464b3e8c00e1e284f037b63350c952ce2d5f641690e3a9a1f
+size 19989692
diff --git a/Keen-qwen4b-20b/checkpoint-13449/model.safetensors b/Keen-qwen4b-20b/checkpoint-13449/model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..f3e62abe75d59aecce1663e7409265d0b94d1015
--- /dev/null
+++ b/Keen-qwen4b-20b/checkpoint-13449/model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5fe83ee3b190b8772bef64da62b031b09516d3666f23deaa2fb32b10416050a6
+size 8411554440
diff --git a/Keen-qwen4b-20b/checkpoint-13449/tokenizer.json b/Keen-qwen4b-20b/checkpoint-13449/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..ec1bde77c388a3d5523769ad975c655ae8ed5eda
--- /dev/null
+++ b/Keen-qwen4b-20b/checkpoint-13449/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4f8a9ce416e1039464b3e8c00e1e284f037b63350c952ce2d5f641690e3a9a1f
+size 19989692
diff --git a/Keen-qwen4b-20b/checkpoint-15698/model.safetensors b/Keen-qwen4b-20b/checkpoint-15698/model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..f922927cb4b97c6fe90104727c759875abe2e36f
--- /dev/null
+++ b/Keen-qwen4b-20b/checkpoint-15698/model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e19c76755aa3b2ce6c0cdf0d170f8d35df339a20e07fb25c9d82cb1d7d22786c
+size 8411554440
diff --git a/Keen-qwen4b-20b/checkpoint-15698/tokenizer.json b/Keen-qwen4b-20b/checkpoint-15698/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..ec1bde77c388a3d5523769ad975c655ae8ed5eda
--- /dev/null
+++ b/Keen-qwen4b-20b/checkpoint-15698/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4f8a9ce416e1039464b3e8c00e1e284f037b63350c952ce2d5f641690e3a9a1f
+size 19989692
diff --git a/Keen-qwen4b-20b/checkpoint-17953/model.safetensors b/Keen-qwen4b-20b/checkpoint-17953/model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..cbf646570eb997c8bdeca573f8bb9731c5d9f78a
--- /dev/null
+++ b/Keen-qwen4b-20b/checkpoint-17953/model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:701a90729ccc31bada5fc4dff6e90a8ac7b90afbbb85f8f80816d693e65dc296
+size 8411554440
diff --git a/Keen-qwen4b-20b/checkpoint-17953/tokenizer.json b/Keen-qwen4b-20b/checkpoint-17953/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..ec1bde77c388a3d5523769ad975c655ae8ed5eda
--- /dev/null
+++ b/Keen-qwen4b-20b/checkpoint-17953/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4f8a9ce416e1039464b3e8c00e1e284f037b63350c952ce2d5f641690e3a9a1f
+size 19989692
diff --git a/Keen-qwen4b-20b/checkpoint-20218/model.safetensors b/Keen-qwen4b-20b/checkpoint-20218/model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..d0fb8cf7684f5e3966f4f051546c4de6018588fb
--- /dev/null
+++ b/Keen-qwen4b-20b/checkpoint-20218/model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:afdc1e1f396a76fe3cabfac49d643dda0e1019743e5350f8a4e67cd16e668866
+size 8411554440
diff --git a/Keen-qwen4b-20b/checkpoint-20218/tokenizer.json b/Keen-qwen4b-20b/checkpoint-20218/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..ec1bde77c388a3d5523769ad975c655ae8ed5eda
--- /dev/null
+++ b/Keen-qwen4b-20b/checkpoint-20218/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4f8a9ce416e1039464b3e8c00e1e284f037b63350c952ce2d5f641690e3a9a1f
+size 19989692
diff --git a/Keen-qwen4b-20b/checkpoint-22478/model.safetensors b/Keen-qwen4b-20b/checkpoint-22478/model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..383472cf997e3f6b1c0d1c0a377198612070c20b
--- /dev/null
+++ b/Keen-qwen4b-20b/checkpoint-22478/model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b869467d5db824e6df4a995bfaca0ae74a7b8e43899593b877b774bcc3861811
+size 8411554440
diff --git a/Keen-qwen4b-20b/checkpoint-22478/tokenizer.json b/Keen-qwen4b-20b/checkpoint-22478/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..ec1bde77c388a3d5523769ad975c655ae8ed5eda
--- /dev/null
+++ b/Keen-qwen4b-20b/checkpoint-22478/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4f8a9ce416e1039464b3e8c00e1e284f037b63350c952ce2d5f641690e3a9a1f
+size 19989692
diff --git a/Keen-qwen4b-20b/checkpoint-2253/model.safetensors b/Keen-qwen4b-20b/checkpoint-2253/model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..bfcc30fbfc47f625c44737c47e57e09c632cb567
--- /dev/null
+++ b/Keen-qwen4b-20b/checkpoint-2253/model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f7816a1cae0b08b1e0901c378b916df2769b9b7fa7ebfac63e98854c33b656a4
+size 8411554440
diff --git a/Keen-qwen4b-20b/checkpoint-2253/tokenizer.json b/Keen-qwen4b-20b/checkpoint-2253/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..ec1bde77c388a3d5523769ad975c655ae8ed5eda
--- /dev/null
+++ b/Keen-qwen4b-20b/checkpoint-2253/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4f8a9ce416e1039464b3e8c00e1e284f037b63350c952ce2d5f641690e3a9a1f
+size 19989692
diff --git a/Keen-qwen4b-20b/checkpoint-4453/model.safetensors b/Keen-qwen4b-20b/checkpoint-4453/model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..3fab30bb7027f20c83ebe18637e0bd52f46b32f4
--- /dev/null
+++ b/Keen-qwen4b-20b/checkpoint-4453/model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9bcfa8a9a42d9b1362d339b13e364fe04ae52c9460f58a715901f3cac64c1a44
+size 8411554440
diff --git a/Keen-qwen4b-20b/checkpoint-4453/tokenizer.json b/Keen-qwen4b-20b/checkpoint-4453/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..ec1bde77c388a3d5523769ad975c655ae8ed5eda
--- /dev/null
+++ b/Keen-qwen4b-20b/checkpoint-4453/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4f8a9ce416e1039464b3e8c00e1e284f037b63350c952ce2d5f641690e3a9a1f
+size 19989692
diff --git a/Keen-qwen4b-20b/checkpoint-6701/model.safetensors b/Keen-qwen4b-20b/checkpoint-6701/model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..79a325f60754a347e6db74dcd8c445dcfdf25e86
--- /dev/null
+++ b/Keen-qwen4b-20b/checkpoint-6701/model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:94e659545b1ba9eecb5c9a2f03206e893bd544f9b4388e591b4316cabf7f1120
+size 8411554440
diff --git a/Keen-qwen4b-20b/checkpoint-6701/tokenizer.json b/Keen-qwen4b-20b/checkpoint-6701/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..ec1bde77c388a3d5523769ad975c655ae8ed5eda
--- /dev/null
+++ b/Keen-qwen4b-20b/checkpoint-6701/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4f8a9ce416e1039464b3e8c00e1e284f037b63350c952ce2d5f641690e3a9a1f
+size 19989692
diff --git a/Keen-qwen4b-20b/checkpoint-8956/model.safetensors b/Keen-qwen4b-20b/checkpoint-8956/model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..441504d1b58ebd1dbde1cdd4528fb2712e74c2da
--- /dev/null
+++ b/Keen-qwen4b-20b/checkpoint-8956/model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8d54e7eb42dffdfc89613913ba3a6c7bfcf0e9eb66ad193c978608eb652efc4a
+size 8411554440
diff --git a/Keen-qwen4b-20b/checkpoint-8956/tokenizer.json b/Keen-qwen4b-20b/checkpoint-8956/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..ec1bde77c388a3d5523769ad975c655ae8ed5eda
--- /dev/null
+++ b/Keen-qwen4b-20b/checkpoint-8956/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4f8a9ce416e1039464b3e8c00e1e284f037b63350c952ce2d5f641690e3a9a1f
+size 19989692
diff --git a/Vanilla-qwen4b-20b/checkpoint-11452/README.md b/Vanilla-qwen4b-20b/checkpoint-11452/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..f8e4ca3eef9ea76dac89627746d5153e7056b124
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-11452/README.md
@@ -0,0 +1,40 @@
+# Vanilla-qwen4b-20b: checkpoint-11452
+
+Text-only Qwen3.5-4B architecture, trained from random initialization.
+This is a base pretraining checkpoint, not an instruction-tuned chat model.
+Training used 4,096-token packs with document-isolated attention.
+The larger positional limit in the architecture configuration does not imply
+validated long-context capability beyond the training context.
+
+- Optimizer step: 11452
+- Cumulative body tokens: 12,000,591,962
+- Parameters: 4,205,751,296
+- Precision: BF16
+- Special document-end token: `<|endoftext|>`
+
+Vanilla uses ordinary passage continuation without a facts prefix.
+
+Use Transformers 5.6.2 (the training/evaluation version) with its native
+`Qwen3_5ForCausalLM` implementation. Install PyTorch and Accelerate for GPU loading.
+No remote custom model code is required. Use plain text prompts rather than a
+chat template. Inference caching is enabled in the released configuration;
+weights and tokenizer vocabulary are unchanged.
+
+```python
+import torch
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+repo = "meihant/AnonymousModel-mht"
+subfolder = "Vanilla-qwen4b-20b/checkpoint-11452"
+tokenizer = AutoTokenizer.from_pretrained(repo, subfolder=subfolder)
+model = AutoModelForCausalLM.from_pretrained(
+    repo, subfolder=subfolder, dtype=torch.bfloat16,
+    device_map="auto", attn_implementation="eager",
+).eval()
+inputs = tokenizer("Paris", return_tensors="pt").to(model.device)
+with torch.inference_mode():
+    output = model.generate(**inputs, max_new_tokens=64, do_sample=False,
+                            use_cache=True, pad_token_id=tokenizer.eos_token_id)
+print(tokenizer.decode(output[0], skip_special_tokens=False))
+```
+
diff --git a/Vanilla-qwen4b-20b/checkpoint-11452/config.json b/Vanilla-qwen4b-20b/checkpoint-11452/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..65f95a09d021431a51836127668a8ecbc2fc21e2
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-11452/config.json
@@ -0,0 +1,83 @@
+{
+  "architectures": [
+    "Qwen3_5ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "attn_output_gate": true,
+  "bos_token_id": null,
+  "dtype": "bfloat16",
+  "eos_token_id": 248044,
+  "full_attention_interval": 4,
+  "head_dim": 256,
+  "hidden_act": "silu",
+  "hidden_size": 2560,
+  "initializer_range": 0.02,
+  "intermediate_size": 9216,
+  "layer_types": [
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention"
+  ],
+  "linear_conv_kernel_dim": 4,
+  "linear_key_head_dim": 128,
+  "linear_num_key_heads": 16,
+  "linear_num_value_heads": 32,
+  "linear_value_head_dim": 128,
+  "mamba_ssm_dtype": "float32",
+  "max_position_embeddings": 262144,
+  "mlp_only_layers": [],
+  "model_type": "qwen3_5_text",
+  "mtp_num_hidden_layers": 1,
+  "mtp_use_dedicated_embeddings": false,
+  "num_attention_heads": 16,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 4,
+  "pad_token_id": 248044,
+  "partial_rotary_factor": 0.25,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "mrope_interleaved": true,
+    "mrope_section": [
+      11,
+      11,
+      10
+    ],
+    "partial_rotary_factor": 0.25,
+    "rope_theta": 10000000,
+    "rope_type": "default"
+  },
+  "tie_word_embeddings": true,
+  "transformers_version": "5.6.2",
+  "use_cache": true,
+  "vocab_size": 248320
+}
diff --git a/Vanilla-qwen4b-20b/checkpoint-11452/generation_config.json b/Vanilla-qwen4b-20b/checkpoint-11452/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..a81ab53f51ff08c3cf4355dce5dd63e5368908c0
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-11452/generation_config.json
@@ -0,0 +1,9 @@
+{
+  "_from_model_config": true,
+  "eos_token_id": 248044,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 248044,
+  "transformers_version": "5.6.2",
+  "use_cache": true
+}
diff --git a/Vanilla-qwen4b-20b/checkpoint-11452/model.safetensors b/Vanilla-qwen4b-20b/checkpoint-11452/model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..90444a535e618b7bf00dc074cd5ca04745a940c5
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-11452/model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a7009776caa28cd43561b57da0d4e9f3538296df42ffa2807f86a729172c20df
+size 8411554440
diff --git a/Vanilla-qwen4b-20b/checkpoint-11452/tokenizer.json b/Vanilla-qwen4b-20b/checkpoint-11452/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..ec1bde77c388a3d5523769ad975c655ae8ed5eda
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-11452/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4f8a9ce416e1039464b3e8c00e1e284f037b63350c952ce2d5f641690e3a9a1f
+size 19989692
diff --git a/Vanilla-qwen4b-20b/checkpoint-11452/tokenizer_config.json b/Vanilla-qwen4b-20b/checkpoint-11452/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..0a18dea722099b09bef6fc796d7bced132cb93d0
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-11452/tokenizer_config.json
@@ -0,0 +1,47 @@
+{
+  "add_prefix_space": false,
+  "audio_bos_token": "<|audio_start|>",
+  "audio_eos_token": "<|audio_end|>",
+  "audio_token": "<|audio_pad|>",
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|endoftext|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>",
+    "<fact>",
+    "</fact>"
+  ],
+  "image_token": "<|image_pad|>",
+  "model_max_length": 32768,
+  "model_specific_special_tokens": {
+    "audio_bos_token": "<|audio_start|>",
+    "audio_eos_token": "<|audio_end|>",
+    "audio_token": "<|audio_pad|>",
+    "image_token": "<|image_pad|>",
+    "video_token": "<|video_pad|>",
+    "vision_bos_token": "<|vision_start|>",
+    "vision_eos_token": "<|vision_end|>"
+  },
+  "pad_token": "<|endoftext|>",
+  "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null,
+  "video_token": "<|video_pad|>",
+  "vision_bos_token": "<|vision_start|>",
+  "vision_eos_token": "<|vision_end|>"
+}
diff --git a/Vanilla-qwen4b-20b/checkpoint-13362/README.md b/Vanilla-qwen4b-20b/checkpoint-13362/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..aa02b54585a6c461c11cf1c8128f06c1162751e0
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-13362/README.md
@@ -0,0 +1,40 @@
+# Vanilla-qwen4b-20b: checkpoint-13362
+
+Text-only Qwen3.5-4B architecture, trained from random initialization.
+This is a base pretraining checkpoint, not an instruction-tuned chat model.
+Training used 4,096-token packs with document-isolated attention.
+The larger positional limit in the architecture configuration does not imply
+validated long-context capability beyond the training context.
+
+- Optimizer step: 13362
+- Cumulative body tokens: 14,000,166,403
+- Parameters: 4,205,751,296
+- Precision: BF16
+- Special document-end token: `<|endoftext|>`
+
+Vanilla uses ordinary passage continuation without a facts prefix.
+
+Use Transformers 5.6.2 (the training/evaluation version) with its native
+`Qwen3_5ForCausalLM` implementation. Install PyTorch and Accelerate for GPU loading.
+No remote custom model code is required. Use plain text prompts rather than a
+chat template. Inference caching is enabled in the released configuration;
+weights and tokenizer vocabulary are unchanged.
+
+```python
+import torch
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+repo = "meihant/AnonymousModel-mht"
+subfolder = "Vanilla-qwen4b-20b/checkpoint-13362"
+tokenizer = AutoTokenizer.from_pretrained(repo, subfolder=subfolder)
+model = AutoModelForCausalLM.from_pretrained(
+    repo, subfolder=subfolder, dtype=torch.bfloat16,
+    device_map="auto", attn_implementation="eager",
+).eval()
+inputs = tokenizer("Paris", return_tensors="pt").to(model.device)
+with torch.inference_mode():
+    output = model.generate(**inputs, max_new_tokens=64, do_sample=False,
+                            use_cache=True, pad_token_id=tokenizer.eos_token_id)
+print(tokenizer.decode(output[0], skip_special_tokens=False))
+```
+
diff --git a/Vanilla-qwen4b-20b/checkpoint-13362/config.json b/Vanilla-qwen4b-20b/checkpoint-13362/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..65f95a09d021431a51836127668a8ecbc2fc21e2
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-13362/config.json
@@ -0,0 +1,83 @@
+{
+  "architectures": [
+    "Qwen3_5ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "attn_output_gate": true,
+  "bos_token_id": null,
+  "dtype": "bfloat16",
+  "eos_token_id": 248044,
+  "full_attention_interval": 4,
+  "head_dim": 256,
+  "hidden_act": "silu",
+  "hidden_size": 2560,
+  "initializer_range": 0.02,
+  "intermediate_size": 9216,
+  "layer_types": [
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention"
+  ],
+  "linear_conv_kernel_dim": 4,
+  "linear_key_head_dim": 128,
+  "linear_num_key_heads": 16,
+  "linear_num_value_heads": 32,
+  "linear_value_head_dim": 128,
+  "mamba_ssm_dtype": "float32",
+  "max_position_embeddings": 262144,
+  "mlp_only_layers": [],
+  "model_type": "qwen3_5_text",
+  "mtp_num_hidden_layers": 1,
+  "mtp_use_dedicated_embeddings": false,
+  "num_attention_heads": 16,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 4,
+  "pad_token_id": 248044,
+  "partial_rotary_factor": 0.25,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "mrope_interleaved": true,
+    "mrope_section": [
+      11,
+      11,
+      10
+    ],
+    "partial_rotary_factor": 0.25,
+    "rope_theta": 10000000,
+    "rope_type": "default"
+  },
+  "tie_word_embeddings": true,
+  "transformers_version": "5.6.2",
+  "use_cache": true,
+  "vocab_size": 248320
+}
diff --git a/Vanilla-qwen4b-20b/checkpoint-13362/generation_config.json b/Vanilla-qwen4b-20b/checkpoint-13362/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..a81ab53f51ff08c3cf4355dce5dd63e5368908c0
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-13362/generation_config.json
@@ -0,0 +1,9 @@
+{
+  "_from_model_config": true,
+  "eos_token_id": 248044,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 248044,
+  "transformers_version": "5.6.2",
+  "use_cache": true
+}
diff --git a/Vanilla-qwen4b-20b/checkpoint-13362/model.safetensors b/Vanilla-qwen4b-20b/checkpoint-13362/model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..e0dde0393e643ee4a474c51d0f478b8bfd09bf55
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-13362/model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aa377d8aebe815164abafc2dd8a85ccc3e44fadc7552e4cb20014167ca09ee74
+size 8411554440
diff --git a/Vanilla-qwen4b-20b/checkpoint-13362/tokenizer.json b/Vanilla-qwen4b-20b/checkpoint-13362/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..ec1bde77c388a3d5523769ad975c655ae8ed5eda
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-13362/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4f8a9ce416e1039464b3e8c00e1e284f037b63350c952ce2d5f641690e3a9a1f
+size 19989692
diff --git a/Vanilla-qwen4b-20b/checkpoint-13362/tokenizer_config.json b/Vanilla-qwen4b-20b/checkpoint-13362/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..0a18dea722099b09bef6fc796d7bced132cb93d0
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-13362/tokenizer_config.json
@@ -0,0 +1,47 @@
+{
+  "add_prefix_space": false,
+  "audio_bos_token": "<|audio_start|>",
+  "audio_eos_token": "<|audio_end|>",
+  "audio_token": "<|audio_pad|>",
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|endoftext|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>",
+    "<fact>",
+    "</fact>"
+  ],
+  "image_token": "<|image_pad|>",
+  "model_max_length": 32768,
+  "model_specific_special_tokens": {
+    "audio_bos_token": "<|audio_start|>",
+    "audio_eos_token": "<|audio_end|>",
+    "audio_token": "<|audio_pad|>",
+    "image_token": "<|image_pad|>",
+    "video_token": "<|video_pad|>",
+    "vision_bos_token": "<|vision_start|>",
+    "vision_eos_token": "<|vision_end|>"
+  },
+  "pad_token": "<|endoftext|>",
+  "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null,
+  "video_token": "<|video_pad|>",
+  "vision_bos_token": "<|vision_start|>",
+  "vision_eos_token": "<|vision_end|>"
+}
diff --git a/Vanilla-qwen4b-20b/checkpoint-15273/README.md b/Vanilla-qwen4b-20b/checkpoint-15273/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..00a848c9554cab577290520b9d6594e49fcb4538
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-15273/README.md
@@ -0,0 +1,40 @@
+# Vanilla-qwen4b-20b: checkpoint-15273
+
+Text-only Qwen3.5-4B architecture, trained from random initialization.
+This is a base pretraining checkpoint, not an instruction-tuned chat model.
+Training used 4,096-token packs with document-isolated attention.
+The larger positional limit in the architecture configuration does not imply
+validated long-context capability beyond the training context.
+
+- Optimizer step: 15273
+- Cumulative body tokens: 16,000,320,768
+- Parameters: 4,205,751,296
+- Precision: BF16
+- Special document-end token: `<|endoftext|>`
+
+Vanilla uses ordinary passage continuation without a facts prefix.
+
+Use Transformers 5.6.2 (the training/evaluation version) with its native
+`Qwen3_5ForCausalLM` implementation. Install PyTorch and Accelerate for GPU loading.
+No remote custom model code is required. Use plain text prompts rather than a
+chat template. Inference caching is enabled in the released configuration;
+weights and tokenizer vocabulary are unchanged.
+
+```python
+import torch
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+repo = "meihant/AnonymousModel-mht"
+subfolder = "Vanilla-qwen4b-20b/checkpoint-15273"
+tokenizer = AutoTokenizer.from_pretrained(repo, subfolder=subfolder)
+model = AutoModelForCausalLM.from_pretrained(
+    repo, subfolder=subfolder, dtype=torch.bfloat16,
+    device_map="auto", attn_implementation="eager",
+).eval()
+inputs = tokenizer("Paris", return_tensors="pt").to(model.device)
+with torch.inference_mode():
+    output = model.generate(**inputs, max_new_tokens=64, do_sample=False,
+                            use_cache=True, pad_token_id=tokenizer.eos_token_id)
+print(tokenizer.decode(output[0], skip_special_tokens=False))
+```
+
diff --git a/Vanilla-qwen4b-20b/checkpoint-15273/config.json b/Vanilla-qwen4b-20b/checkpoint-15273/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..65f95a09d021431a51836127668a8ecbc2fc21e2
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-15273/config.json
@@ -0,0 +1,83 @@
+{
+  "architectures": [
+    "Qwen3_5ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "attn_output_gate": true,
+  "bos_token_id": null,
+  "dtype": "bfloat16",
+  "eos_token_id": 248044,
+  "full_attention_interval": 4,
+  "head_dim": 256,
+  "hidden_act": "silu",
+  "hidden_size": 2560,
+  "initializer_range": 0.02,
+  "intermediate_size": 9216,
+  "layer_types": [
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention"
+  ],
+  "linear_conv_kernel_dim": 4,
+  "linear_key_head_dim": 128,
+  "linear_num_key_heads": 16,
+  "linear_num_value_heads": 32,
+  "linear_value_head_dim": 128,
+  "mamba_ssm_dtype": "float32",
+  "max_position_embeddings": 262144,
+  "mlp_only_layers": [],
+  "model_type": "qwen3_5_text",
+  "mtp_num_hidden_layers": 1,
+  "mtp_use_dedicated_embeddings": false,
+  "num_attention_heads": 16,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 4,
+  "pad_token_id": 248044,
+  "partial_rotary_factor": 0.25,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "mrope_interleaved": true,
+    "mrope_section": [
+      11,
+      11,
+      10
+    ],
+    "partial_rotary_factor": 0.25,
+    "rope_theta": 10000000,
+    "rope_type": "default"
+  },
+  "tie_word_embeddings": true,
+  "transformers_version": "5.6.2",
+  "use_cache": true,
+  "vocab_size": 248320
+}
diff --git a/Vanilla-qwen4b-20b/checkpoint-15273/generation_config.json b/Vanilla-qwen4b-20b/checkpoint-15273/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..a81ab53f51ff08c3cf4355dce5dd63e5368908c0
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-15273/generation_config.json
@@ -0,0 +1,9 @@
+{
+  "_from_model_config": true,
+  "eos_token_id": 248044,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 248044,
+  "transformers_version": "5.6.2",
+  "use_cache": true
+}
diff --git a/Vanilla-qwen4b-20b/checkpoint-15273/model.safetensors b/Vanilla-qwen4b-20b/checkpoint-15273/model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..069035078dac3bf03c87f31a49c825467dc92829
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-15273/model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9233f915f89cd02a49cbd8690c5de7625f483dcc697cc2a451137522613e464a
+size 8411554440
diff --git a/Vanilla-qwen4b-20b/checkpoint-15273/tokenizer.json b/Vanilla-qwen4b-20b/checkpoint-15273/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..ec1bde77c388a3d5523769ad975c655ae8ed5eda
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-15273/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4f8a9ce416e1039464b3e8c00e1e284f037b63350c952ce2d5f641690e3a9a1f
+size 19989692
diff --git a/Vanilla-qwen4b-20b/checkpoint-15273/tokenizer_config.json b/Vanilla-qwen4b-20b/checkpoint-15273/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..0a18dea722099b09bef6fc796d7bced132cb93d0
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-15273/tokenizer_config.json
@@ -0,0 +1,47 @@
+{
+  "add_prefix_space": false,
+  "audio_bos_token": "<|audio_start|>",
+  "audio_eos_token": "<|audio_end|>",
+  "audio_token": "<|audio_pad|>",
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|endoftext|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>",
+    "<fact>",
+    "</fact>"
+  ],
+  "image_token": "<|image_pad|>",
+  "model_max_length": 32768,
+  "model_specific_special_tokens": {
+    "audio_bos_token": "<|audio_start|>",
+    "audio_eos_token": "<|audio_end|>",
+    "audio_token": "<|audio_pad|>",
+    "image_token": "<|image_pad|>",
+    "video_token": "<|video_pad|>",
+    "vision_bos_token": "<|vision_start|>",
+    "vision_eos_token": "<|vision_end|>"
+  },
+  "pad_token": "<|endoftext|>",
+  "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null,
+  "video_token": "<|video_pad|>",
+  "vision_bos_token": "<|vision_start|>",
+  "vision_eos_token": "<|vision_end|>"
+}
diff --git a/Vanilla-qwen4b-20b/checkpoint-17188/README.md b/Vanilla-qwen4b-20b/checkpoint-17188/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..2816ca9368f4cd9c30ac249dd7ac206fe3117932
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-17188/README.md
@@ -0,0 +1,40 @@
+# Vanilla-qwen4b-20b: checkpoint-17188
+
+Text-only Qwen3.5-4B architecture, trained from random initialization.
+This is a base pretraining checkpoint, not an instruction-tuned chat model.
+Training used 4,096-token packs with document-isolated attention.
+The larger positional limit in the architecture configuration does not imply
+validated long-context capability beyond the training context.
+
+- Optimizer step: 17188
+- Cumulative body tokens: 18,000,257,314
+- Parameters: 4,205,751,296
+- Precision: BF16
+- Special document-end token: `<|endoftext|>`
+
+Vanilla uses ordinary passage continuation without a facts prefix.
+
+Use Transformers 5.6.2 (the training/evaluation version) with its native
+`Qwen3_5ForCausalLM` implementation. Install PyTorch and Accelerate for GPU loading.
+No remote custom model code is required. Use plain text prompts rather than a
+chat template. Inference caching is enabled in the released configuration;
+weights and tokenizer vocabulary are unchanged.
+
+```python
+import torch
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+repo = "meihant/AnonymousModel-mht"
+subfolder = "Vanilla-qwen4b-20b/checkpoint-17188"
+tokenizer = AutoTokenizer.from_pretrained(repo, subfolder=subfolder)
+model = AutoModelForCausalLM.from_pretrained(
+    repo, subfolder=subfolder, dtype=torch.bfloat16,
+    device_map="auto", attn_implementation="eager",
+).eval()
+inputs = tokenizer("Paris", return_tensors="pt").to(model.device)
+with torch.inference_mode():
+    output = model.generate(**inputs, max_new_tokens=64, do_sample=False,
+                            use_cache=True, pad_token_id=tokenizer.eos_token_id)
+print(tokenizer.decode(output[0], skip_special_tokens=False))
+```
+
diff --git a/Vanilla-qwen4b-20b/checkpoint-17188/config.json b/Vanilla-qwen4b-20b/checkpoint-17188/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..65f95a09d021431a51836127668a8ecbc2fc21e2
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-17188/config.json
@@ -0,0 +1,83 @@
+{
+  "architectures": [
+    "Qwen3_5ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "attn_output_gate": true,
+  "bos_token_id": null,
+  "dtype": "bfloat16",
+  "eos_token_id": 248044,
+  "full_attention_interval": 4,
+  "head_dim": 256,
+  "hidden_act": "silu",
+  "hidden_size": 2560,
+  "initializer_range": 0.02,
+  "intermediate_size": 9216,
+  "layer_types": [
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention"
+  ],
+  "linear_conv_kernel_dim": 4,
+  "linear_key_head_dim": 128,
+  "linear_num_key_heads": 16,
+  "linear_num_value_heads": 32,
+  "linear_value_head_dim": 128,
+  "mamba_ssm_dtype": "float32",
+  "max_position_embeddings": 262144,
+  "mlp_only_layers": [],
+  "model_type": "qwen3_5_text",
+  "mtp_num_hidden_layers": 1,
+  "mtp_use_dedicated_embeddings": false,
+  "num_attention_heads": 16,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 4,
+  "pad_token_id": 248044,
+  "partial_rotary_factor": 0.25,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "mrope_interleaved": true,
+    "mrope_section": [
+      11,
+      11,
+      10
+    ],
+    "partial_rotary_factor": 0.25,
+    "rope_theta": 10000000,
+    "rope_type": "default"
+  },
+  "tie_word_embeddings": true,
+  "transformers_version": "5.6.2",
+  "use_cache": true,
+  "vocab_size": 248320
+}
diff --git a/Vanilla-qwen4b-20b/checkpoint-17188/generation_config.json b/Vanilla-qwen4b-20b/checkpoint-17188/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..a81ab53f51ff08c3cf4355dce5dd63e5368908c0
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-17188/generation_config.json
@@ -0,0 +1,9 @@
+{
+  "_from_model_config": true,
+  "eos_token_id": 248044,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 248044,
+  "transformers_version": "5.6.2",
+  "use_cache": true
+}
diff --git a/Vanilla-qwen4b-20b/checkpoint-17188/model.safetensors b/Vanilla-qwen4b-20b/checkpoint-17188/model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..4571321c6df4fd45efbff7fdf3936699700b8596
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-17188/model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:67d469d450411055d454f462e379f69c74b39a771ef14ffc569589a61978117d
+size 8411554440
diff --git a/Vanilla-qwen4b-20b/checkpoint-17188/tokenizer.json b/Vanilla-qwen4b-20b/checkpoint-17188/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..ec1bde77c388a3d5523769ad975c655ae8ed5eda
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-17188/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4f8a9ce416e1039464b3e8c00e1e284f037b63350c952ce2d5f641690e3a9a1f
+size 19989692
diff --git a/Vanilla-qwen4b-20b/checkpoint-17188/tokenizer_config.json b/Vanilla-qwen4b-20b/checkpoint-17188/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..0a18dea722099b09bef6fc796d7bced132cb93d0
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-17188/tokenizer_config.json
@@ -0,0 +1,47 @@
+{
+  "add_prefix_space": false,
+  "audio_bos_token": "<|audio_start|>",
+  "audio_eos_token": "<|audio_end|>",
+  "audio_token": "<|audio_pad|>",
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|endoftext|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>",
+    "<fact>",
+    "</fact>"
+  ],
+  "image_token": "<|image_pad|>",
+  "model_max_length": 32768,
+  "model_specific_special_tokens": {
+    "audio_bos_token": "<|audio_start|>",
+    "audio_eos_token": "<|audio_end|>",
+    "audio_token": "<|audio_pad|>",
+    "image_token": "<|image_pad|>",
+    "video_token": "<|video_pad|>",
+    "vision_bos_token": "<|vision_start|>",
+    "vision_eos_token": "<|vision_end|>"
+  },
+  "pad_token": "<|endoftext|>",
+  "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null,
+  "video_token": "<|video_pad|>",
+  "vision_bos_token": "<|vision_start|>",
+  "vision_eos_token": "<|vision_end|>"
+}
diff --git a/Vanilla-qwen4b-20b/checkpoint-1910/config.json b/Vanilla-qwen4b-20b/checkpoint-1910/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..65f95a09d021431a51836127668a8ecbc2fc21e2
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-1910/config.json
@@ -0,0 +1,83 @@
+{
+  "architectures": [
+    "Qwen3_5ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "attn_output_gate": true,
+  "bos_token_id": null,
+  "dtype": "bfloat16",
+  "eos_token_id": 248044,
+  "full_attention_interval": 4,
+  "head_dim": 256,
+  "hidden_act": "silu",
+  "hidden_size": 2560,
+  "initializer_range": 0.02,
+  "intermediate_size": 9216,
+  "layer_types": [
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention"
+  ],
+  "linear_conv_kernel_dim": 4,
+  "linear_key_head_dim": 128,
+  "linear_num_key_heads": 16,
+  "linear_num_value_heads": 32,
+  "linear_value_head_dim": 128,
+  "mamba_ssm_dtype": "float32",
+  "max_position_embeddings": 262144,
+  "mlp_only_layers": [],
+  "model_type": "qwen3_5_text",
+  "mtp_num_hidden_layers": 1,
+  "mtp_use_dedicated_embeddings": false,
+  "num_attention_heads": 16,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 4,
+  "pad_token_id": 248044,
+  "partial_rotary_factor": 0.25,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "mrope_interleaved": true,
+    "mrope_section": [
+      11,
+      11,
+      10
+    ],
+    "partial_rotary_factor": 0.25,
+    "rope_theta": 10000000,
+    "rope_type": "default"
+  },
+  "tie_word_embeddings": true,
+  "transformers_version": "5.6.2",
+  "use_cache": true,
+  "vocab_size": 248320
+}
diff --git a/Vanilla-qwen4b-20b/checkpoint-1910/generation_config.json b/Vanilla-qwen4b-20b/checkpoint-1910/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..a81ab53f51ff08c3cf4355dce5dd63e5368908c0
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-1910/generation_config.json
@@ -0,0 +1,9 @@
+{
+  "_from_model_config": true,
+  "eos_token_id": 248044,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 248044,
+  "transformers_version": "5.6.2",
+  "use_cache": true
+}
diff --git a/Vanilla-qwen4b-20b/checkpoint-1910/model.safetensors b/Vanilla-qwen4b-20b/checkpoint-1910/model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..d1c07051ed91d04dcb90a1b955d082ccebc70cc0
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-1910/model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ebbecc5de3ed774dbfce3387c20cba60244884b3690bf9aa59c26a11769e5557
+size 8411554440
diff --git a/Vanilla-qwen4b-20b/checkpoint-1910/tokenizer.json b/Vanilla-qwen4b-20b/checkpoint-1910/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..ec1bde77c388a3d5523769ad975c655ae8ed5eda
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-1910/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4f8a9ce416e1039464b3e8c00e1e284f037b63350c952ce2d5f641690e3a9a1f
+size 19989692
diff --git a/Vanilla-qwen4b-20b/checkpoint-1910/tokenizer_config.json b/Vanilla-qwen4b-20b/checkpoint-1910/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..0a18dea722099b09bef6fc796d7bced132cb93d0
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-1910/tokenizer_config.json
@@ -0,0 +1,47 @@
+{
+  "add_prefix_space": false,
+  "audio_bos_token": "<|audio_start|>",
+  "audio_eos_token": "<|audio_end|>",
+  "audio_token": "<|audio_pad|>",
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|endoftext|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>",
+    "<fact>",
+    "</fact>"
+  ],
+  "image_token": "<|image_pad|>",
+  "model_max_length": 32768,
+  "model_specific_special_tokens": {
+    "audio_bos_token": "<|audio_start|>",
+    "audio_eos_token": "<|audio_end|>",
+    "audio_token": "<|audio_pad|>",
+    "image_token": "<|image_pad|>",
+    "video_token": "<|video_pad|>",
+    "vision_bos_token": "<|vision_start|>",
+    "vision_eos_token": "<|vision_end|>"
+  },
+  "pad_token": "<|endoftext|>",
+  "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null,
+  "video_token": "<|video_pad|>",
+  "vision_bos_token": "<|vision_start|>",
+  "vision_eos_token": "<|vision_end|>"
+}
diff --git a/Vanilla-qwen4b-20b/checkpoint-19106/README.md b/Vanilla-qwen4b-20b/checkpoint-19106/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..a218ec80bf0a7977ad877c0e6d370f3306312616
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-19106/README.md
@@ -0,0 +1,40 @@
+# Vanilla-qwen4b-20b: checkpoint-19106
+
+Text-only Qwen3.5-4B architecture, trained from random initialization.
+This is a base pretraining checkpoint, not an instruction-tuned chat model.
+Training used 4,096-token packs with document-isolated attention.
+The larger positional limit in the architecture configuration does not imply
+validated long-context capability beyond the training context.
+
+- Optimizer step: 19106
+- Cumulative body tokens: 20,000,220,345
+- Parameters: 4,205,751,296
+- Precision: BF16
+- Special document-end token: `<|endoftext|>`
+
+Vanilla uses ordinary passage continuation without a facts prefix.
+
+Use Transformers 5.6.2 (the training/evaluation version) with its native
+`Qwen3_5ForCausalLM` implementation. Install PyTorch and Accelerate for GPU loading.
+No remote custom model code is required. Use plain text prompts rather than a
+chat template. Inference caching is enabled in the released configuration;
+weights and tokenizer vocabulary are unchanged.
+
+```python
+import torch
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+repo = "meihant/AnonymousModel-mht"
+subfolder = "Vanilla-qwen4b-20b/checkpoint-19106"
+tokenizer = AutoTokenizer.from_pretrained(repo, subfolder=subfolder)
+model = AutoModelForCausalLM.from_pretrained(
+    repo, subfolder=subfolder, dtype=torch.bfloat16,
+    device_map="auto", attn_implementation="eager",
+).eval()
+inputs = tokenizer("Paris", return_tensors="pt").to(model.device)
+with torch.inference_mode():
+    output = model.generate(**inputs, max_new_tokens=64, do_sample=False,
+                            use_cache=True, pad_token_id=tokenizer.eos_token_id)
+print(tokenizer.decode(output[0], skip_special_tokens=False))
+```
+
diff --git a/Vanilla-qwen4b-20b/checkpoint-19106/config.json b/Vanilla-qwen4b-20b/checkpoint-19106/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..65f95a09d021431a51836127668a8ecbc2fc21e2
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-19106/config.json
@@ -0,0 +1,83 @@
+{
+  "architectures": [
+    "Qwen3_5ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "attn_output_gate": true,
+  "bos_token_id": null,
+  "dtype": "bfloat16",
+  "eos_token_id": 248044,
+  "full_attention_interval": 4,
+  "head_dim": 256,
+  "hidden_act": "silu",
+  "hidden_size": 2560,
+  "initializer_range": 0.02,
+  "intermediate_size": 9216,
+  "layer_types": [
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention"
+  ],
+  "linear_conv_kernel_dim": 4,
+  "linear_key_head_dim": 128,
+  "linear_num_key_heads": 16,
+  "linear_num_value_heads": 32,
+  "linear_value_head_dim": 128,
+  "mamba_ssm_dtype": "float32",
+  "max_position_embeddings": 262144,
+  "mlp_only_layers": [],
+  "model_type": "qwen3_5_text",
+  "mtp_num_hidden_layers": 1,
+  "mtp_use_dedicated_embeddings": false,
+  "num_attention_heads": 16,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 4,
+  "pad_token_id": 248044,
+  "partial_rotary_factor": 0.25,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "mrope_interleaved": true,
+    "mrope_section": [
+      11,
+      11,
+      10
+    ],
+    "partial_rotary_factor": 0.25,
+    "rope_theta": 10000000,
+    "rope_type": "default"
+  },
+  "tie_word_embeddings": true,
+  "transformers_version": "5.6.2",
+  "use_cache": true,
+  "vocab_size": 248320
+}
diff --git a/Vanilla-qwen4b-20b/checkpoint-19106/generation_config.json b/Vanilla-qwen4b-20b/checkpoint-19106/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..a81ab53f51ff08c3cf4355dce5dd63e5368908c0
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-19106/generation_config.json
@@ -0,0 +1,9 @@
+{
+  "_from_model_config": true,
+  "eos_token_id": 248044,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 248044,
+  "transformers_version": "5.6.2",
+  "use_cache": true
+}
diff --git a/Vanilla-qwen4b-20b/checkpoint-19106/model.safetensors b/Vanilla-qwen4b-20b/checkpoint-19106/model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..c0f712dfb80cb444b0095b71a6adb82bcaa660d8
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-19106/model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8ae4543af3abe122675c0252d7bc0d1f8fcfcac8292e5064a9d824a2a75eb8e3
+size 8411554440
diff --git a/Vanilla-qwen4b-20b/checkpoint-19106/tokenizer.json b/Vanilla-qwen4b-20b/checkpoint-19106/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..ec1bde77c388a3d5523769ad975c655ae8ed5eda
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-19106/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4f8a9ce416e1039464b3e8c00e1e284f037b63350c952ce2d5f641690e3a9a1f
+size 19989692
diff --git a/Vanilla-qwen4b-20b/checkpoint-19106/tokenizer_config.json b/Vanilla-qwen4b-20b/checkpoint-19106/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..0a18dea722099b09bef6fc796d7bced132cb93d0
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-19106/tokenizer_config.json
@@ -0,0 +1,47 @@
+{
+  "add_prefix_space": false,
+  "audio_bos_token": "<|audio_start|>",
+  "audio_eos_token": "<|audio_end|>",
+  "audio_token": "<|audio_pad|>",
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|endoftext|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>",
+    "<fact>",
+    "</fact>"
+  ],
+  "image_token": "<|image_pad|>",
+  "model_max_length": 32768,
+  "model_specific_special_tokens": {
+    "audio_bos_token": "<|audio_start|>",
+    "audio_eos_token": "<|audio_end|>",
+    "audio_token": "<|audio_pad|>",
+    "image_token": "<|image_pad|>",
+    "video_token": "<|video_pad|>",
+    "vision_bos_token": "<|vision_start|>",
+    "vision_eos_token": "<|vision_end|>"
+  },
+  "pad_token": "<|endoftext|>",
+  "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null,
+  "video_token": "<|video_pad|>",
+  "vision_bos_token": "<|vision_start|>",
+  "vision_eos_token": "<|vision_end|>"
+}
diff --git a/Vanilla-qwen4b-20b/checkpoint-3818/README.md b/Vanilla-qwen4b-20b/checkpoint-3818/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..ace0d4dd1a27366175a10aeeda4fb8a2ddb7861b
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-3818/README.md
@@ -0,0 +1,40 @@
+# Vanilla-qwen4b-20b: checkpoint-3818
+
+Text-only Qwen3.5-4B architecture, trained from random initialization.
+This is a base pretraining checkpoint, not an instruction-tuned chat model.
+Training used 4,096-token packs with document-isolated attention.
+The larger positional limit in the architecture configuration does not imply
+validated long-context capability beyond the training context.
+
+- Optimizer step: 3818
+- Cumulative body tokens: 4,000,211,289
+- Parameters: 4,205,751,296
+- Precision: BF16
+- Special document-end token: `<|endoftext|>`
+
+Vanilla uses ordinary passage continuation without a facts prefix.
+
+Use Transformers 5.6.2 (the training/evaluation version) with its native
+`Qwen3_5ForCausalLM` implementation. Install PyTorch and Accelerate for GPU loading.
+No remote custom model code is required. Use plain text prompts rather than a
+chat template. Inference caching is enabled in the released configuration;
+weights and tokenizer vocabulary are unchanged.
+
+```python
+import torch
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+repo = "meihant/AnonymousModel-mht"
+subfolder = "Vanilla-qwen4b-20b/checkpoint-3818"
+tokenizer = AutoTokenizer.from_pretrained(repo, subfolder=subfolder)
+model = AutoModelForCausalLM.from_pretrained(
+    repo, subfolder=subfolder, dtype=torch.bfloat16,
+    device_map="auto", attn_implementation="eager",
+).eval()
+inputs = tokenizer("Paris", return_tensors="pt").to(model.device)
+with torch.inference_mode():
+    output = model.generate(**inputs, max_new_tokens=64, do_sample=False,
+                            use_cache=True, pad_token_id=tokenizer.eos_token_id)
+print(tokenizer.decode(output[0], skip_special_tokens=False))
+```
+
diff --git a/Vanilla-qwen4b-20b/checkpoint-3818/config.json b/Vanilla-qwen4b-20b/checkpoint-3818/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..65f95a09d021431a51836127668a8ecbc2fc21e2
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-3818/config.json
@@ -0,0 +1,83 @@
+{
+  "architectures": [
+    "Qwen3_5ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "attn_output_gate": true,
+  "bos_token_id": null,
+  "dtype": "bfloat16",
+  "eos_token_id": 248044,
+  "full_attention_interval": 4,
+  "head_dim": 256,
+  "hidden_act": "silu",
+  "hidden_size": 2560,
+  "initializer_range": 0.02,
+  "intermediate_size": 9216,
+  "layer_types": [
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention"
+  ],
+  "linear_conv_kernel_dim": 4,
+  "linear_key_head_dim": 128,
+  "linear_num_key_heads": 16,
+  "linear_num_value_heads": 32,
+  "linear_value_head_dim": 128,
+  "mamba_ssm_dtype": "float32",
+  "max_position_embeddings": 262144,
+  "mlp_only_layers": [],
+  "model_type": "qwen3_5_text",
+  "mtp_num_hidden_layers": 1,
+  "mtp_use_dedicated_embeddings": false,
+  "num_attention_heads": 16,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 4,
+  "pad_token_id": 248044,
+  "partial_rotary_factor": 0.25,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "mrope_interleaved": true,
+    "mrope_section": [
+      11,
+      11,
+      10
+    ],
+    "partial_rotary_factor": 0.25,
+    "rope_theta": 10000000,
+    "rope_type": "default"
+  },
+  "tie_word_embeddings": true,
+  "transformers_version": "5.6.2",
+  "use_cache": true,
+  "vocab_size": 248320
+}
diff --git a/Vanilla-qwen4b-20b/checkpoint-3818/generation_config.json b/Vanilla-qwen4b-20b/checkpoint-3818/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..a81ab53f51ff08c3cf4355dce5dd63e5368908c0
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-3818/generation_config.json
@@ -0,0 +1,9 @@
+{
+  "_from_model_config": true,
+  "eos_token_id": 248044,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 248044,
+  "transformers_version": "5.6.2",
+  "use_cache": true
+}
diff --git a/Vanilla-qwen4b-20b/checkpoint-3818/model.safetensors b/Vanilla-qwen4b-20b/checkpoint-3818/model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..bff56189ed7221e3958d3125b7329eafecfdf8c2
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-3818/model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4f00f81c6fc90de35737f7e36e1ac56e4cc381b3dba06da29b60a0474f91fa21
+size 8411554440
diff --git a/Vanilla-qwen4b-20b/checkpoint-3818/tokenizer.json b/Vanilla-qwen4b-20b/checkpoint-3818/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..ec1bde77c388a3d5523769ad975c655ae8ed5eda
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-3818/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4f8a9ce416e1039464b3e8c00e1e284f037b63350c952ce2d5f641690e3a9a1f
+size 19989692
diff --git a/Vanilla-qwen4b-20b/checkpoint-3818/tokenizer_config.json b/Vanilla-qwen4b-20b/checkpoint-3818/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..0a18dea722099b09bef6fc796d7bced132cb93d0
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-3818/tokenizer_config.json
@@ -0,0 +1,47 @@
+{
+  "add_prefix_space": false,
+  "audio_bos_token": "<|audio_start|>",
+  "audio_eos_token": "<|audio_end|>",
+  "audio_token": "<|audio_pad|>",
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|endoftext|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>",
+    "<fact>",
+    "</fact>"
+  ],
+  "image_token": "<|image_pad|>",
+  "model_max_length": 32768,
+  "model_specific_special_tokens": {
+    "audio_bos_token": "<|audio_start|>",
+    "audio_eos_token": "<|audio_end|>",
+    "audio_token": "<|audio_pad|>",
+    "image_token": "<|image_pad|>",
+    "video_token": "<|video_pad|>",
+    "vision_bos_token": "<|vision_start|>",
+    "vision_eos_token": "<|vision_end|>"
+  },
+  "pad_token": "<|endoftext|>",
+  "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null,
+  "video_token": "<|video_pad|>",
+  "vision_bos_token": "<|vision_start|>",
+  "vision_eos_token": "<|vision_end|>"
+}
diff --git a/Vanilla-qwen4b-20b/checkpoint-5726/README.md b/Vanilla-qwen4b-20b/checkpoint-5726/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..fb06711892fd62aeeab4c7b83d237dfdcee8f101
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-5726/README.md
@@ -0,0 +1,40 @@
+# Vanilla-qwen4b-20b: checkpoint-5726
+
+Text-only Qwen3.5-4B architecture, trained from random initialization.
+This is a base pretraining checkpoint, not an instruction-tuned chat model.
+Training used 4,096-token packs with document-isolated attention.
+The larger positional limit in the architecture configuration does not imply
+validated long-context capability beyond the training context.
+
+- Optimizer step: 5726
+- Cumulative body tokens: 6,000,159,072
+- Parameters: 4,205,751,296
+- Precision: BF16
+- Special document-end token: `<|endoftext|>`
+
+Vanilla uses ordinary passage continuation without a facts prefix.
+
+Use Transformers 5.6.2 (the training/evaluation version) with its native
+`Qwen3_5ForCausalLM` implementation. Install PyTorch and Accelerate for GPU loading.
+No remote custom model code is required. Use plain text prompts rather than a
+chat template. Inference caching is enabled in the released configuration;
+weights and tokenizer vocabulary are unchanged.
+
+```python
+import torch
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+repo = "meihant/AnonymousModel-mht"
+subfolder = "Vanilla-qwen4b-20b/checkpoint-5726"
+tokenizer = AutoTokenizer.from_pretrained(repo, subfolder=subfolder)
+model = AutoModelForCausalLM.from_pretrained(
+    repo, subfolder=subfolder, dtype=torch.bfloat16,
+    device_map="auto", attn_implementation="eager",
+).eval()
+inputs = tokenizer("Paris", return_tensors="pt").to(model.device)
+with torch.inference_mode():
+    output = model.generate(**inputs, max_new_tokens=64, do_sample=False,
+                            use_cache=True, pad_token_id=tokenizer.eos_token_id)
+print(tokenizer.decode(output[0], skip_special_tokens=False))
+```
+
diff --git a/Vanilla-qwen4b-20b/checkpoint-5726/config.json b/Vanilla-qwen4b-20b/checkpoint-5726/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..65f95a09d021431a51836127668a8ecbc2fc21e2
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-5726/config.json
@@ -0,0 +1,83 @@
+{
+  "architectures": [
+    "Qwen3_5ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "attn_output_gate": true,
+  "bos_token_id": null,
+  "dtype": "bfloat16",
+  "eos_token_id": 248044,
+  "full_attention_interval": 4,
+  "head_dim": 256,
+  "hidden_act": "silu",
+  "hidden_size": 2560,
+  "initializer_range": 0.02,
+  "intermediate_size": 9216,
+  "layer_types": [
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention"
+  ],
+  "linear_conv_kernel_dim": 4,
+  "linear_key_head_dim": 128,
+  "linear_num_key_heads": 16,
+  "linear_num_value_heads": 32,
+  "linear_value_head_dim": 128,
+  "mamba_ssm_dtype": "float32",
+  "max_position_embeddings": 262144,
+  "mlp_only_layers": [],
+  "model_type": "qwen3_5_text",
+  "mtp_num_hidden_layers": 1,
+  "mtp_use_dedicated_embeddings": false,
+  "num_attention_heads": 16,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 4,
+  "pad_token_id": 248044,
+  "partial_rotary_factor": 0.25,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "mrope_interleaved": true,
+    "mrope_section": [
+      11,
+      11,
+      10
+    ],
+    "partial_rotary_factor": 0.25,
+    "rope_theta": 10000000,
+    "rope_type": "default"
+  },
+  "tie_word_embeddings": true,
+  "transformers_version": "5.6.2",
+  "use_cache": true,
+  "vocab_size": 248320
+}
diff --git a/Vanilla-qwen4b-20b/checkpoint-5726/generation_config.json b/Vanilla-qwen4b-20b/checkpoint-5726/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..a81ab53f51ff08c3cf4355dce5dd63e5368908c0
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-5726/generation_config.json
@@ -0,0 +1,9 @@
+{
+  "_from_model_config": true,
+  "eos_token_id": 248044,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 248044,
+  "transformers_version": "5.6.2",
+  "use_cache": true
+}
diff --git a/Vanilla-qwen4b-20b/checkpoint-5726/model.safetensors b/Vanilla-qwen4b-20b/checkpoint-5726/model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..fd69b18d9320976f83445b1b9cecea8959de932d
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-5726/model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:67da1a0d5464184db38d0612ed09cee7e38c962bd0fc15063caf52ee11b55997
+size 8411554440
diff --git a/Vanilla-qwen4b-20b/checkpoint-5726/tokenizer.json b/Vanilla-qwen4b-20b/checkpoint-5726/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..ec1bde77c388a3d5523769ad975c655ae8ed5eda
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-5726/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4f8a9ce416e1039464b3e8c00e1e284f037b63350c952ce2d5f641690e3a9a1f
+size 19989692
diff --git a/Vanilla-qwen4b-20b/checkpoint-5726/tokenizer_config.json b/Vanilla-qwen4b-20b/checkpoint-5726/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..0a18dea722099b09bef6fc796d7bced132cb93d0
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-5726/tokenizer_config.json
@@ -0,0 +1,47 @@
+{
+  "add_prefix_space": false,
+  "audio_bos_token": "<|audio_start|>",
+  "audio_eos_token": "<|audio_end|>",
+  "audio_token": "<|audio_pad|>",
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|endoftext|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>",
+    "<fact>",
+    "</fact>"
+  ],
+  "image_token": "<|image_pad|>",
+  "model_max_length": 32768,
+  "model_specific_special_tokens": {
+    "audio_bos_token": "<|audio_start|>",
+    "audio_eos_token": "<|audio_end|>",
+    "audio_token": "<|audio_pad|>",
+    "image_token": "<|image_pad|>",
+    "video_token": "<|video_pad|>",
+    "vision_bos_token": "<|vision_start|>",
+    "vision_eos_token": "<|vision_end|>"
+  },
+  "pad_token": "<|endoftext|>",
+  "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null,
+  "video_token": "<|video_pad|>",
+  "vision_bos_token": "<|vision_start|>",
+  "vision_eos_token": "<|vision_end|>"
+}
diff --git a/Vanilla-qwen4b-20b/checkpoint-7635/README.md b/Vanilla-qwen4b-20b/checkpoint-7635/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..0649a25ad9e5744b9b86cf32ffbffae93ea0c8b7
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-7635/README.md
@@ -0,0 +1,40 @@
+# Vanilla-qwen4b-20b: checkpoint-7635
+
+Text-only Qwen3.5-4B architecture, trained from random initialization.
+This is a base pretraining checkpoint, not an instruction-tuned chat model.
+Training used 4,096-token packs with document-isolated attention.
+The larger positional limit in the architecture configuration does not imply
+validated long-context capability beyond the training context.
+
+- Optimizer step: 7635
+- Cumulative body tokens: 8,000,913,248
+- Parameters: 4,205,751,296
+- Precision: BF16
+- Special document-end token: `<|endoftext|>`
+
+Vanilla uses ordinary passage continuation without a facts prefix.
+
+Use Transformers 5.6.2 (the training/evaluation version) with its native
+`Qwen3_5ForCausalLM` implementation. Install PyTorch and Accelerate for GPU loading.
+No remote custom model code is required. Use plain text prompts rather than a
+chat template. Inference caching is enabled in the released configuration;
+weights and tokenizer vocabulary are unchanged.
+
+```python
+import torch
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+repo = "meihant/AnonymousModel-mht"
+subfolder = "Vanilla-qwen4b-20b/checkpoint-7635"
+tokenizer = AutoTokenizer.from_pretrained(repo, subfolder=subfolder)
+model = AutoModelForCausalLM.from_pretrained(
+    repo, subfolder=subfolder, dtype=torch.bfloat16,
+    device_map="auto", attn_implementation="eager",
+).eval()
+inputs = tokenizer("Paris", return_tensors="pt").to(model.device)
+with torch.inference_mode():
+    output = model.generate(**inputs, max_new_tokens=64, do_sample=False,
+                            use_cache=True, pad_token_id=tokenizer.eos_token_id)
+print(tokenizer.decode(output[0], skip_special_tokens=False))
+```
+
diff --git a/Vanilla-qwen4b-20b/checkpoint-7635/config.json b/Vanilla-qwen4b-20b/checkpoint-7635/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..65f95a09d021431a51836127668a8ecbc2fc21e2
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-7635/config.json
@@ -0,0 +1,83 @@
+{
+  "architectures": [
+    "Qwen3_5ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "attn_output_gate": true,
+  "bos_token_id": null,
+  "dtype": "bfloat16",
+  "eos_token_id": 248044,
+  "full_attention_interval": 4,
+  "head_dim": 256,
+  "hidden_act": "silu",
+  "hidden_size": 2560,
+  "initializer_range": 0.02,
+  "intermediate_size": 9216,
+  "layer_types": [
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention",
+    "linear_attention",
+    "linear_attention",
+    "linear_attention",
+    "full_attention"
+  ],
+  "linear_conv_kernel_dim": 4,
+  "linear_key_head_dim": 128,
+  "linear_num_key_heads": 16,
+  "linear_num_value_heads": 32,
+  "linear_value_head_dim": 128,
+  "mamba_ssm_dtype": "float32",
+  "max_position_embeddings": 262144,
+  "mlp_only_layers": [],
+  "model_type": "qwen3_5_text",
+  "mtp_num_hidden_layers": 1,
+  "mtp_use_dedicated_embeddings": false,
+  "num_attention_heads": 16,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 4,
+  "pad_token_id": 248044,
+  "partial_rotary_factor": 0.25,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "mrope_interleaved": true,
+    "mrope_section": [
+      11,
+      11,
+      10
+    ],
+    "partial_rotary_factor": 0.25,
+    "rope_theta": 10000000,
+    "rope_type": "default"
+  },
+  "tie_word_embeddings": true,
+  "transformers_version": "5.6.2",
+  "use_cache": true,
+  "vocab_size": 248320
+}
diff --git a/Vanilla-qwen4b-20b/checkpoint-7635/generation_config.json b/Vanilla-qwen4b-20b/checkpoint-7635/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..a81ab53f51ff08c3cf4355dce5dd63e5368908c0
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-7635/generation_config.json
@@ -0,0 +1,9 @@
+{
+  "_from_model_config": true,
+  "eos_token_id": 248044,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 248044,
+  "transformers_version": "5.6.2",
+  "use_cache": true
+}
diff --git a/Vanilla-qwen4b-20b/checkpoint-7635/model.safetensors b/Vanilla-qwen4b-20b/checkpoint-7635/model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..97913d43075fa8bd25b22b23c04d6c019786d797
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-7635/model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b718e6cf6d61bfa6e864e85d6bee37079a3f4f030c420311885ca8b4297a1c03
+size 8411554440
diff --git a/Vanilla-qwen4b-20b/checkpoint-7635/tokenizer.json b/Vanilla-qwen4b-20b/checkpoint-7635/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..ec1bde77c388a3d5523769ad975c655ae8ed5eda
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-7635/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4f8a9ce416e1039464b3e8c00e1e284f037b63350c952ce2d5f641690e3a9a1f
+size 19989692
diff --git a/Vanilla-qwen4b-20b/checkpoint-7635/tokenizer_config.json b/Vanilla-qwen4b-20b/checkpoint-7635/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..0a18dea722099b09bef6fc796d7bced132cb93d0
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-7635/tokenizer_config.json
@@ -0,0 +1,47 @@
+{
+  "add_prefix_space": false,
+  "audio_bos_token": "<|audio_start|>",
+  "audio_eos_token": "<|audio_end|>",
+  "audio_token": "<|audio_pad|>",
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|endoftext|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>",
+    "<fact>",
+    "</fact>"
+  ],
+  "image_token": "<|image_pad|>",
+  "model_max_length": 32768,
+  "model_specific_special_tokens": {
+    "audio_bos_token": "<|audio_start|>",
+    "audio_eos_token": "<|audio_end|>",
+    "audio_token": "<|audio_pad|>",
+    "image_token": "<|image_pad|>",
+    "video_token": "<|video_pad|>",
+    "vision_bos_token": "<|vision_start|>",
+    "vision_eos_token": "<|vision_end|>"
+  },
+  "pad_token": "<|endoftext|>",
+  "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null,
+  "video_token": "<|video_pad|>",
+  "vision_bos_token": "<|vision_start|>",
+  "vision_eos_token": "<|vision_end|>"
+}
diff --git a/Vanilla-qwen4b-20b/checkpoint-9543/model.safetensors b/Vanilla-qwen4b-20b/checkpoint-9543/model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..f7288f4952f972006867cf83b6b30ceff22ac540
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-9543/model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7bb6c5406dcb41f68006815bf25e67a42320b08c78d822b7ac4bd9c29fd3a13a
+size 8411554440
diff --git a/Vanilla-qwen4b-20b/checkpoint-9543/tokenizer.json b/Vanilla-qwen4b-20b/checkpoint-9543/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..ec1bde77c388a3d5523769ad975c655ae8ed5eda
--- /dev/null
+++ b/Vanilla-qwen4b-20b/checkpoint-9543/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4f8a9ce416e1039464b3e8c00e1e284f037b63350c952ce2d5f641690e3a9a1f
+size 19989692