Update README.md

README.md

@@ -5,20 +5,18 @@ language:
 license: apache-2.0
 base_model: facebook/bart-large
 tags:
-- generated_from_trainer
-model-index:
-- name: bart-large-summary-map-reduce-1024
-  results: []
+- map-reduce
+- summarization
+datasets:
+- pszemraj/summary-map-reduce
+pipeline_tag: text2text-generation
 ---
 
-<!-- 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. -->
-
-# bart-large-summary-map-reduce-1024
+# bart-large-summary-map-reduce
 
 A text2text model to "map-reduce" summaries of a chunked long document into one.
 
-An explanation of this model's role: 
+An explanation of this model's role as a post-processor for [textsum](https://github.com/pszemraj/textsum) (_or any other long-doc summarization method similar to the below_)
 
 ![image/png](https://cdn-uploads.huggingface.co/production/uploads/60bccec062080d33f875cd0c/Sv7_-MM901qNkyHuBdTC_.png)
 
@@ -39,11 +37,9 @@ an example of aggregating summaries from chunks of a long document:
 import torch
 from transformers import pipeline
 
-model_name = "pszemraj/bart-large-summary-map-reduce-1024"
-
 pipe = pipeline(
     "text2text-generation",
-    model=model_name,
+    model="pszemraj/bart-large-summary-map-reduce",
     device_map="auto",
 )
 
@@ -58,10 +54,11 @@ text = """A computer implemented method of generating a syntactic object. The me
     The brain is constantly loosing neurons because you doesn&#39;t want all the junk around."""
 
 # generate
-torch.cuda.empty_cache()
+if torch.cuda.is_available():
+    torch.cuda.empty_cache()
 res = pipe(
     text,
-    max_new_tokens=512,
+    max_new_tokens=512, # increase up to 1024 if needed
     num_beams=4,
     early_stopping=True,
     truncation=True,
@@ -83,4 +80,4 @@ The following hyperparameters were used during training:
 - optimizer: Use OptimizerNames.PAGED_ADAMW with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
 - lr_scheduler_type: cosine
 - lr_scheduler_warmup_ratio: 0.05
-- num_epochs: 3.0
+- num_epochs: 3.0