Rename Stage C model package artifact

MODEL_CARD.md

@@ -1,6 +1,6 @@
 # TMCRA TokenGraph-LLM Stage C Model Card
 
-This model card mirrors the Hugging Face README. The current recommended artifact is `tgclm_stagec_model_package_20260606.zip`, which contains the Stage C checkpoint, tokenizer, dataset manifest, public training summary, checksum, and evaluation notes.
+This model card mirrors the Hugging Face README. The current recommended artifact is `tmcra_tokengraph_stagec_model_package_20260606.zip`, which contains the Stage C checkpoint, tokenizer, dataset manifest, public training summary, checksum, and evaluation notes.
 
 Source code and full-chain training scripts:
 

README.md

@@ -23,7 +23,7 @@ This Hugging Face repository hosts model artifacts. Full source code, training s
 Current default checkpoint:
 
 - release line: `v0.2.0-stagec`
-- package: `tgclm_stagec_model_package_20260606.zip`
+- package: `tmcra_tokengraph_stagec_model_package_20260606.zip`
 - checkpoint inside package: `checkpoint/token_graph_dynamic_decoder_v3.pt`
 - parameters: `114,615,372`
 - shape: `dim=512`, `graph_layers=8`, `decoder_layers=10`
@@ -31,7 +31,7 @@ Current default checkpoint:
 - precision during training: `bf16`
 - effective training samples: about `1.03M`
 - training steps: `62,000`
-- SHA256: `1cebd34362e51a2e5c3e5ba8e96df04b353b763f10b73acc13a67442a3b2dae3`
+- SHA256: `cc23285628eaed47c20009b6be6b5eb0600ded57ac2e09519370d97158fecd33`
 
 Legacy v0.1 files may still be present in this repository for historical comparison. The Stage C package is the current recommended artifact.
 

README_zh.md

@@ -9,7 +9,7 @@ https://github.com/reshuibuduo/TMCRA-TokenGraph-LLM
 ## 当前模型
 
 - 版本线：`v0.2.0-stagec`
-- 模型包：`tgclm_stagec_model_package_20260606.zip`
+- 模型包：`tmcra_tokengraph_stagec_model_package_20260606.zip`
 - checkpoint：`checkpoint/token_graph_dynamic_decoder_v3.pt`
 - 参数量：`114,615,372`
 - 结构：`dim=512`，`graph_layers=8`，`decoder_layers=10`
@@ -17,7 +17,7 @@ https://github.com/reshuibuduo/TMCRA-TokenGraph-LLM
 - 训练精度：`bf16`
 - 有效训练样本：约 `1.03M`
 - 训练步数：`62,000`
-- SHA256：`1cebd34362e51a2e5c3e5ba8e96df04b353b763f10b73acc13a67442a3b2dae3`
+- SHA256：`cc23285628eaed47c20009b6be6b5eb0600ded57ac2e09519370d97158fecd33`
 
 旧 v0.1 文件可能仍保留用于历史对比。当前推荐使用 Stage C 模型包。
 

SHA256SUMS.txt

@@ -1 +1 @@
-1cebd34362e51a2e5c3e5ba8e96df04b353b763f10b73acc13a67442a3b2dae3  tgclm_stagec_model_package_20260606.zip
+cc23285628eaed47c20009b6be6b5eb0600ded57ac2e09519370d97158fecd33  tmcra_tokengraph_stagec_model_package_20260606.zip

tmcra_tokengraph_stagec_model_package_20260606.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cc23285628eaed47c20009b6be6b5eb0600ded57ac2e09519370d97158fecd33
+size 426173844