Training in progress, step 500

config.json

@@ -1,5 +1,5 @@
 {
-  "_name_or_path": "google/long-t5-tglobal-base",
+  "_name_or_path": "google/long-t5-local-base",
   "architectures": [
     "LongT5ForConditionalGeneration"
   ],
@@ -9,7 +9,7 @@
   "decoder_start_token_id": 0,
   "dense_act_fn": "gelu_new",
   "dropout_rate": 0.1,
-  "encoder_attention_type": "transient-global",
+  "encoder_attention_type": "local",
   "eos_token_id": 1,
   "feed_forward_proj": "gated-gelu",
   "global_block_size": 16,

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:859f5bc5f3cdd9c9ed5ba58f15d34c155e500d96410fa00955309a82add11865
-size 1187780840
+oid sha256:ffbb2d180ce7c221565431f76418e1baf9543670c95600b0bdc82805187fa25c
+size 1089042632

runs/Oct02_16-04-42_179dde1d5d6e/events.out.tfevents.1727885083.179dde1d5d6e.558.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f9d64273d6140f54c404d2abae8620d0af70e368c69c1ffa0c792e05c2f14248
+size 9012

runs/Oct02_16-39-30_179dde1d5d6e/events.out.tfevents.1727887170.179dde1d5d6e.558.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:13e897f107a80e5bd389c11f1621bec9dd8409f9c60b0b43b7fa525d4e348a07
+size 5502

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4ed359f12127ec5e02fefbf21c892088890378706bfa9781ef8106a87c6e5091
-size 5368
+oid sha256:b1b8d7cc23a5e821af29c7753478a713ca77f13adfef3a6b1e38783183f0448a
+size 5304