Commit
•
2837990
1
Parent(s):
6b3de4b
Training in progress, step 5000
Browse files- .gitignore +1 -0
- config.json +62 -0
- pytorch_model.bin +3 -0
- runs/Feb01_20-10-23_5059af6dd429/1675282230.4192584/events.out.tfevents.1675282230.5059af6dd429.40843.1 +3 -0
- runs/Feb01_20-10-23_5059af6dd429/1675283292.4863887/events.out.tfevents.1675283292.5059af6dd429.40843.3 +3 -0
- runs/Feb01_20-10-23_5059af6dd429/1675283697.7299592/events.out.tfevents.1675283697.5059af6dd429.40843.5 +3 -0
- runs/Feb01_20-10-23_5059af6dd429/1675285249.347183/events.out.tfevents.1675285249.5059af6dd429.40843.7 +3 -0
- runs/Feb01_20-10-23_5059af6dd429/events.out.tfevents.1675282230.5059af6dd429.40843.0 +3 -0
- runs/Feb01_20-10-23_5059af6dd429/events.out.tfevents.1675283292.5059af6dd429.40843.2 +3 -0
- runs/Feb01_20-10-23_5059af6dd429/events.out.tfevents.1675283697.5059af6dd429.40843.4 +3 -0
- runs/Feb01_20-10-23_5059af6dd429/events.out.tfevents.1675285249.5059af6dd429.40843.6 +3 -0
- runs/Feb01_21-54-22_5059af6dd429/1675288468.0510192/events.out.tfevents.1675288468.5059af6dd429.72103.1 +3 -0
- runs/Feb01_21-54-22_5059af6dd429/1675289991.3852987/events.out.tfevents.1675289991.5059af6dd429.72103.2 +3 -0
- runs/Feb01_21-54-22_5059af6dd429/1675290006.0077693/events.out.tfevents.1675290006.5059af6dd429.72103.4 +3 -0
- runs/Feb01_21-54-22_5059af6dd429/events.out.tfevents.1675288468.5059af6dd429.72103.0 +3 -0
- runs/Feb01_21-54-22_5059af6dd429/events.out.tfevents.1675290006.5059af6dd429.72103.3 +3 -0
- training_args.bin +3 -0
.gitignore
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
checkpoint-*/
|
config.json
ADDED
@@ -0,0 +1,62 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "sberbank-ai/ruT5-base",
|
3 |
+
"_num_labels": 2,
|
4 |
+
"architectures": [
|
5 |
+
"T5ForConditionalGeneration"
|
6 |
+
],
|
7 |
+
"d_ff": 3072,
|
8 |
+
"d_kv": 64,
|
9 |
+
"d_model": 768,
|
10 |
+
"decoder_start_token_id": 0,
|
11 |
+
"dense_act_fn": "relu",
|
12 |
+
"dropout_rate": 0.1,
|
13 |
+
"eos_token_id": 2,
|
14 |
+
"feed_forward_proj": "relu",
|
15 |
+
"initializer_factor": 1.0,
|
16 |
+
"is_encoder_decoder": true,
|
17 |
+
"is_gated_act": false,
|
18 |
+
"layer_norm_epsilon": 1e-06,
|
19 |
+
"max_length": 256,
|
20 |
+
"model_type": "t5",
|
21 |
+
"n_positions": 512,
|
22 |
+
"num_decoder_layers": 12,
|
23 |
+
"num_heads": 12,
|
24 |
+
"num_layers": 12,
|
25 |
+
"output_past": true,
|
26 |
+
"pad_token_id": 0,
|
27 |
+
"relative_attention_max_distance": 128,
|
28 |
+
"relative_attention_num_buckets": 32,
|
29 |
+
"task_specific_params": {
|
30 |
+
"summarization": {
|
31 |
+
"early_stopping": true,
|
32 |
+
"length_penalty": 2.0,
|
33 |
+
"max_length": 200,
|
34 |
+
"min_length": 30,
|
35 |
+
"no_repeat_ngram_size": 3,
|
36 |
+
"num_beams": 4,
|
37 |
+
"prefix": "summarize: "
|
38 |
+
},
|
39 |
+
"translation_en_to_de": {
|
40 |
+
"early_stopping": true,
|
41 |
+
"max_length": 300,
|
42 |
+
"num_beams": 4,
|
43 |
+
"prefix": "translate English to German: "
|
44 |
+
},
|
45 |
+
"translation_en_to_fr": {
|
46 |
+
"early_stopping": true,
|
47 |
+
"max_length": 300,
|
48 |
+
"num_beams": 4,
|
49 |
+
"prefix": "translate English to French: "
|
50 |
+
},
|
51 |
+
"translation_en_to_ro": {
|
52 |
+
"early_stopping": true,
|
53 |
+
"max_length": 300,
|
54 |
+
"num_beams": 4,
|
55 |
+
"prefix": "translate English to Romanian: "
|
56 |
+
}
|
57 |
+
},
|
58 |
+
"torch_dtype": "float32",
|
59 |
+
"transformers_version": "4.26.0",
|
60 |
+
"use_cache": true,
|
61 |
+
"vocab_size": 32100
|
62 |
+
}
|
pytorch_model.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:908484f80f1b6317e5f03362c1d773b4442f754c969293434fc353f946961fe1
|
3 |
+
size 891616913
|
runs/Feb01_20-10-23_5059af6dd429/1675282230.4192584/events.out.tfevents.1675282230.5059af6dd429.40843.1
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:adb0a479b9f54f4767a32c3818e9610e9e18304d3e9c80009190ef90bbc0011c
|
3 |
+
size 5891
|
runs/Feb01_20-10-23_5059af6dd429/1675283292.4863887/events.out.tfevents.1675283292.5059af6dd429.40843.3
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:4672aa5ad12c346d20a7af59ba41b808af607bea01cb0ebedfd59e5487b10195
|
3 |
+
size 5891
|
runs/Feb01_20-10-23_5059af6dd429/1675283697.7299592/events.out.tfevents.1675283697.5059af6dd429.40843.5
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:c27deadc609c22add14d04fb634e8eb2e38ccb3cad0d0a659c53db2626ae2216
|
3 |
+
size 5891
|
runs/Feb01_20-10-23_5059af6dd429/1675285249.347183/events.out.tfevents.1675285249.5059af6dd429.40843.7
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:94a839970767fd3b0ffe6655c4aeb829b965939216eb988f97e97afee9f91fe1
|
3 |
+
size 5891
|
runs/Feb01_20-10-23_5059af6dd429/events.out.tfevents.1675282230.5059af6dd429.40843.0
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:9274dcec4582ea5d061adb92feb753028a50c6f32e62cb627f4a7b6d4a6d3653
|
3 |
+
size 33009
|
runs/Feb01_20-10-23_5059af6dd429/events.out.tfevents.1675283292.5059af6dd429.40843.2
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:cee9f2814cc4b9588f6717cca66a78d9da5012627f09f069d6a62e3e18d6b6bf
|
3 |
+
size 15631
|
runs/Feb01_20-10-23_5059af6dd429/events.out.tfevents.1675283697.5059af6dd429.40843.4
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:315a5037a43afead0b5c720932b654947743986fe83db8cdd23c83aa5ee9886b
|
3 |
+
size 46094
|
runs/Feb01_20-10-23_5059af6dd429/events.out.tfevents.1675285249.5059af6dd429.40843.6
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:fe92d700850bf89b7932078529a62db878cb7f78e0159226b34e0f3bf80d81e8
|
3 |
+
size 75355
|
runs/Feb01_21-54-22_5059af6dd429/1675288468.0510192/events.out.tfevents.1675288468.5059af6dd429.72103.1
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:acc1f7a09aa68f08efd246e86472c3789920f0c741c581f2cb9f5a69b8047654
|
3 |
+
size 5891
|
runs/Feb01_21-54-22_5059af6dd429/1675289991.3852987/events.out.tfevents.1675289991.5059af6dd429.72103.2
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d27a71cbb1c56363ee9af718e8ccfdf65817990705f76abce27e13a1e235d8ac
|
3 |
+
size 5891
|
runs/Feb01_21-54-22_5059af6dd429/1675290006.0077693/events.out.tfevents.1675290006.5059af6dd429.72103.4
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:bb4510e4d40016a6ec7be8aef3c8aa3fd31921d5545341e7ec0e60888327731d
|
3 |
+
size 5891
|
runs/Feb01_21-54-22_5059af6dd429/events.out.tfevents.1675288468.5059af6dd429.72103.0
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:fe8d76c9dcae8672f5d7e710325270b54db7a67f9db7e63b18aedaebb8776dc1
|
3 |
+
size 50834
|
runs/Feb01_21-54-22_5059af6dd429/events.out.tfevents.1675290006.5059af6dd429.72103.3
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:2899b41e148effe54d4534d657f30c5c187dfcdbf81f345a3df8e1ff24f5ffb6
|
3 |
+
size 88494
|
training_args.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:c5319ad6196d6a88bd202594a1d54076fca05c9d377ee393bb866fd18269faed
|
3 |
+
size 3579
|