Training in progress, step 500
Browse files- config.json +6 -6
- pytorch_model.bin +2 -2
- runs/May29_13-19-43_b012beb56a71/events.out.tfevents.1685366409.b012beb56a71.14200.2 +2 -2
- runs/May29_13-29-19_b012beb56a71/1685366971.4179688/events.out.tfevents.1685366971.b012beb56a71.14200.5 +3 -0
- runs/May29_13-29-19_b012beb56a71/events.out.tfevents.1685366971.b012beb56a71.14200.4 +3 -0
- runs/May29_13-30-03_b012beb56a71/1685367008.69498/events.out.tfevents.1685367008.b012beb56a71.14200.7 +3 -0
- runs/May29_13-30-03_b012beb56a71/events.out.tfevents.1685367008.b012beb56a71.14200.6 +3 -0
- runs/May29_13-30-47_b012beb56a71/1685367051.1968637/events.out.tfevents.1685367051.b012beb56a71.14200.9 +3 -0
- runs/May29_13-30-47_b012beb56a71/events.out.tfevents.1685367051.b012beb56a71.14200.8 +3 -0
- runs/May29_13-32-15_b012beb56a71/1685367143.1890602/events.out.tfevents.1685367143.b012beb56a71.19119.1 +3 -0
- runs/May29_13-32-15_b012beb56a71/events.out.tfevents.1685367143.b012beb56a71.19119.0 +3 -0
- runs/May29_13-33-03_b012beb56a71/1685367188.1156273/events.out.tfevents.1685367188.b012beb56a71.19119.3 +3 -0
- runs/May29_13-33-03_b012beb56a71/events.out.tfevents.1685367188.b012beb56a71.19119.2 +3 -0
- runs/May29_13-33-59_b012beb56a71/1685367248.86299/events.out.tfevents.1685367248.b012beb56a71.19759.1 +3 -0
- runs/May29_13-33-59_b012beb56a71/events.out.tfevents.1685367248.b012beb56a71.19759.0 +3 -0
- training_args.bin +1 -1
config.json
CHANGED
|
@@ -1,11 +1,11 @@
|
|
| 1 |
{
|
| 2 |
-
"_name_or_path": "t5-
|
| 3 |
"architectures": [
|
| 4 |
"T5ForConditionalGeneration"
|
| 5 |
],
|
| 6 |
-
"d_ff":
|
| 7 |
"d_kv": 64,
|
| 8 |
-
"d_model":
|
| 9 |
"decoder_start_token_id": 0,
|
| 10 |
"dense_act_fn": "relu",
|
| 11 |
"dropout_rate": 0.1,
|
|
@@ -17,9 +17,9 @@
|
|
| 17 |
"layer_norm_epsilon": 1e-06,
|
| 18 |
"model_type": "t5",
|
| 19 |
"n_positions": 512,
|
| 20 |
-
"num_decoder_layers":
|
| 21 |
-
"num_heads":
|
| 22 |
-
"num_layers":
|
| 23 |
"output_past": true,
|
| 24 |
"pad_token_id": 0,
|
| 25 |
"relative_attention_max_distance": 128,
|
|
|
|
| 1 |
{
|
| 2 |
+
"_name_or_path": "t5-base",
|
| 3 |
"architectures": [
|
| 4 |
"T5ForConditionalGeneration"
|
| 5 |
],
|
| 6 |
+
"d_ff": 3072,
|
| 7 |
"d_kv": 64,
|
| 8 |
+
"d_model": 768,
|
| 9 |
"decoder_start_token_id": 0,
|
| 10 |
"dense_act_fn": "relu",
|
| 11 |
"dropout_rate": 0.1,
|
|
|
|
| 17 |
"layer_norm_epsilon": 1e-06,
|
| 18 |
"model_type": "t5",
|
| 19 |
"n_positions": 512,
|
| 20 |
+
"num_decoder_layers": 12,
|
| 21 |
+
"num_heads": 12,
|
| 22 |
+
"num_layers": 12,
|
| 23 |
"output_past": true,
|
| 24 |
"pad_token_id": 0,
|
| 25 |
"relative_attention_max_distance": 128,
|
pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:951f606c6b41d6511d1e55fa0b1aeccb4c9ad89424b061a3709f1a265164d8b4
|
| 3 |
+
size 891702929
|
runs/May29_13-19-43_b012beb56a71/events.out.tfevents.1685366409.b012beb56a71.14200.2
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:78d7e155b22980a3bb66ba806ee6c08b667ee783d5b0c5ff31a9527122d32b3d
|
| 3 |
+
size 7338
|
runs/May29_13-29-19_b012beb56a71/1685366971.4179688/events.out.tfevents.1685366971.b012beb56a71.14200.5
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:df3e6147f6b0be013d376e84ef27ae6a3b9d000e904e9cf35930f361a381b5c5
|
| 3 |
+
size 6213
|
runs/May29_13-29-19_b012beb56a71/events.out.tfevents.1685366971.b012beb56a71.14200.4
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c769f028787c124318ec5397f508a8456728d372ed0ca36c4722009bbb5f52b4
|
| 3 |
+
size 4986
|
runs/May29_13-30-03_b012beb56a71/1685367008.69498/events.out.tfevents.1685367008.b012beb56a71.14200.7
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5f9afa83334daa150dc115b080f8b4cd6848ad1be68f952883be5ff6676a461c
|
| 3 |
+
size 6213
|
runs/May29_13-30-03_b012beb56a71/events.out.tfevents.1685367008.b012beb56a71.14200.6
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:dd69487bc1c4edf8318ca5f89f04d0d2ebabd5d0fffc0e13bba611547c111dd5
|
| 3 |
+
size 4184
|
runs/May29_13-30-47_b012beb56a71/1685367051.1968637/events.out.tfevents.1685367051.b012beb56a71.14200.9
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a471fccf0975afe41f69e398c1c4f3b48a0a48d0b0642fc3e4b7d353e112214d
|
| 3 |
+
size 6213
|
runs/May29_13-30-47_b012beb56a71/events.out.tfevents.1685367051.b012beb56a71.14200.8
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e11ab8c152098688c6934213c0e8cf746e57dba5068d47ef0c0cae5f929b6efd
|
| 3 |
+
size 4184
|
runs/May29_13-32-15_b012beb56a71/1685367143.1890602/events.out.tfevents.1685367143.b012beb56a71.19119.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:59f6933942fa8662474fde5f18c4023209863d8792a2ed158f4c81b0f6e202c5
|
| 3 |
+
size 6213
|
runs/May29_13-32-15_b012beb56a71/events.out.tfevents.1685367143.b012beb56a71.19119.0
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0e574d30d7842f95b1febde545bde84806f9eb396b49c3a69b881cb972cd1180
|
| 3 |
+
size 4184
|
runs/May29_13-33-03_b012beb56a71/1685367188.1156273/events.out.tfevents.1685367188.b012beb56a71.19119.3
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6ebb13ab1ecb7c2e0ee4d050e21f3d0e3e122fd282aad994d3a5425d01fd3804
|
| 3 |
+
size 6213
|
runs/May29_13-33-03_b012beb56a71/events.out.tfevents.1685367188.b012beb56a71.19119.2
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:41d9b407531105d5e72c60b9d39a3835f7f07646a22b6f8f3894f42a451e1ecf
|
| 3 |
+
size 4184
|
runs/May29_13-33-59_b012beb56a71/1685367248.86299/events.out.tfevents.1685367248.b012beb56a71.19759.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c73438814c7e60661d3824642875cca29b9420737cd8844404f20ca7cc4cfc96
|
| 3 |
+
size 6213
|
runs/May29_13-33-59_b012beb56a71/events.out.tfevents.1685367248.b012beb56a71.19759.0
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f65e4907d82d97604388bc8bb15f60884be5947f19050bcf659fce5415c8ef9c
|
| 3 |
+
size 6616
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4091
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:340789e5c8e4b8c7adda9e77ec2795ed6ed8d9f9dee7343e663289ab44f84ed8
|
| 3 |
size 4091
|