Update README.md

README.md

@@ -12,9 +12,10 @@ model-index:
 <!-- This model card has been generated automatically according to the information the Trainer had access to. You
 should probably proofread and complete it, then remove this comment. -->
 
-# flan-t5-large-stacked-XSUM-1024-WIP-2p8-850-stacked-xsum-1024-evaluated
+# flan-t5-large-stacked-XSUM-1024
+
+This model is a fine-tuned version of [google/flan-t5-large](https://huggingface.co/google/flan-t5-large) on the stacked-summaries/stacked-xsum-1024 dataset.
 
-This model is a fine-tuned version of [pszemraj/flan-t5-large-stacked-XSUM-1024-WIP-2p8-850](https://huggingface.co/pszemraj/flan-t5-large-stacked-XSUM-1024-WIP-2p8-850) on the stacked-summaries/stacked-xsum-1024 dataset.
 It achieves the following results on the evaluation set:
 - eval_loss: 1.3314
 - eval_rouge1: 46.5061
@@ -25,17 +26,28 @@ It achieves the following results on the evaluation set:
 - eval_runtime: 9456.3608
 - eval_samples_per_second: 1.896
 - eval_steps_per_second: 0.119
-- step: 0
 
+> Note that the evaluation set is `stacked-summaries/stacked-xsum-1024` and not `xsum` itself
 ## Model description
 
-More information needed
+This model card presents a model trained on a stacked dataset, which aims to improve summarization by testing the benefits of "task-oriented pretraining." The model is designed to learn how to effectively condense and distill information from text by stacking summaries and separating them into independent concepts. By doing so, the model can learn to identify essential information without simply mimicking the style of the dataset summaries.
 
 ## Intended uses & limitations
 
-More information needed
+- max input length (in tokens): 1024
 
 ## Training and evaluation data
 
-More information needed
+Refer to `stacked-summaries/stacked-xsum-1024`
+
+Trained for approx 3 epochs before ROUGE scores stabilized on most recent run:
+
+
+### scores
+
+![stable-scores](https://i.imgur.com/4tvhHVy.png)
+
+
+### gradients
 
+![gradients](https://i.imgur.com/V6zcmAb.png)