Add architecture config (4,912,992 params)

config.json

@@ -1,16 +1,17 @@
 {
   "architectures": ["CompactLM"],
-  "vocab_size": 12288,
-  "d_model": 256,
-  "n_layers": 4,
-  "n_heads": 4,
-  "head_dim": 64,
-  "ff_dim": 640,
+  "model_type": "compactlm",
+  "vocab_size": 8192,
+  "hidden_size": 224,
+  "intermediate_size": 576,
+  "num_hidden_layers": 6,
+  "num_attention_heads": 7,
+  "num_key_value_heads": 2,
+  "head_dim": 32,
   "context_length": 512,
+  "rms_norm_eps": 1e-6,
+  "rope_theta": 10000.0,
   "tie_word_embeddings": true,
-  "norm_eps": 1e-05,
-  "rope_base": 10000.0,
-  "n_params": 6162688,
-  "dtype": "float32",
-  "note": "Custom LLaMA-style architecture (RoPE + causal SDPA + SwiGLU + RMSNorm pre-norm). Not transformers-native; load via the provided train_compactlm5m.py CompactLM class or a matching custom loader."
-}
+  "torch_dtype": "float32",
+  "num_parameters": 4912992
+}