Rewrite TCN model README

README.md

@@ -4,16 +4,24 @@ tags:
 - finance
 - tcn
 - time-series
-- tabular
-- private-research
+- pytorch
+- mag7
+- private-dataset-trained
 library_name: pytorch
 ---
 
 # MAG7 TCN Sequence Model
 
-Private model repo for the best TCN checkpoint produced by the MAG7 TCN ZeroGPU Trainer Space.
+Public model artifact repo for a Temporal Convolutional Network trained on a private MAG7 daily ML dataset.
 
-The trainer uploads these artifacts after a run:
+The dataset used for training is private and is not distributed with this model:
+
+- Private dataset: `kyLELEng/mag7-ml-daily-dataset-5y`
+- Default target: `target_next_10d_outperform_qqq`
+- Split method: chronological train / validation / test
+- Training hardware: Hugging Face Jobs, `a10g-large`
+
+After training completes, this repo contains:
 
 - `model.pt`
 - `scaler.joblib`
@@ -22,4 +30,4 @@ The trainer uploads these artifacts after a run:
 - `training_log.csv`
 - `test_predictions.csv`
 
-The source dataset is `kyLELEng/mag7-ml-daily-dataset-5y`.
+This model is for research only. It is not investment advice.