Upload TFT5ForConditionalGeneration

README.md

@@ -15,8 +15,8 @@ probably proofread and complete it, then remove this comment. -->
 
 This model is a fine-tuned version of [t5-small](https://huggingface.co/t5-small) on an unknown dataset.
 It achieves the following results on the evaluation set:
-- Train Loss: 1.7547
-- Validation Loss: 1.5748
+- Train Loss: 1.7589
+- Validation Loss: 1.5686
 - Epoch: 0
 
 ## Model description
@@ -43,7 +43,7 @@ The following hyperparameters were used during training:
 
 | Train Loss | Validation Loss | Epoch |
 |:----------:|:---------------:|:-----:|
-| 1.7547     | 1.5748          | 0     |
+| 1.7589     | 1.5686          | 0     |
 
 
 ### Framework versions

config.json

@@ -18,16 +18,16 @@
   "layer_norm_epsilon": 1e-06,
   "model_type": "t5",
   "n_positions": 512,
-  "num_decoder_layers": 12,
-  "num_heads": 12,
-  "num_layers": 12,
+  "num_decoder_layers": 6,
+  "num_heads": 8,
+  "num_layers": 6,
   "output_past": true,
   "pad_token_id": 0,
   "relative_attention_max_distance": 128,
   "relative_attention_num_buckets": 32,
   "task_specific_params": {
     "summarization": {
-      "early_stopping": false,
+      "early_stopping": true,
       "length_penalty": 2.0,
       "max_length": 200,
       "min_length": 30,

tf_model.h5

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f087c9540f2dcad5934501f4ba306ec2906d46e0373eda8b84f5819878313942
+oid sha256:f544b37435c3669d8ab8be538d75012407d9656bd2f62f10d825a7a15c2330ab
 size 373902664