Training in progress, step 500
Browse files- config.json +34 -0
- model.safetensors +3 -0
- runs/Apr12_11-34-18_495dcc39ad78/events.out.tfevents.1744457676.495dcc39ad78.672.0 +3 -0
- runs/Apr12_11-35-38_495dcc39ad78/events.out.tfevents.1744457741.495dcc39ad78.672.1 +3 -0
- runs/Apr12_11-38-15_495dcc39ad78/events.out.tfevents.1744457905.495dcc39ad78.3074.0 +3 -0
- runs/Apr12_11-39-52_495dcc39ad78/events.out.tfevents.1744458014.495dcc39ad78.3873.0 +3 -0
- training_args.bin +3 -0
config.json
ADDED
@@ -0,0 +1,34 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"activation_dropout": 0.0,
|
3 |
+
"activation_function": "relu",
|
4 |
+
"architectures": [
|
5 |
+
"M2M100ForConditionalGeneration"
|
6 |
+
],
|
7 |
+
"attention_dropout": 0.1,
|
8 |
+
"bos_token_id": 0,
|
9 |
+
"d_model": 1024,
|
10 |
+
"decoder_attention_heads": 16,
|
11 |
+
"decoder_ffn_dim": 4096,
|
12 |
+
"decoder_layerdrop": 0,
|
13 |
+
"decoder_layers": 12,
|
14 |
+
"decoder_start_token_id": 2,
|
15 |
+
"dropout": 0.1,
|
16 |
+
"encoder_attention_heads": 16,
|
17 |
+
"encoder_ffn_dim": 4096,
|
18 |
+
"encoder_layerdrop": 0,
|
19 |
+
"encoder_layers": 12,
|
20 |
+
"eos_token_id": 2,
|
21 |
+
"init_std": 0.02,
|
22 |
+
"is_encoder_decoder": true,
|
23 |
+
"max_length": null,
|
24 |
+
"max_position_embeddings": 1024,
|
25 |
+
"model_type": "m2m_100",
|
26 |
+
"num_hidden_layers": 12,
|
27 |
+
"pad_token_id": 1,
|
28 |
+
"scale_embedding": true,
|
29 |
+
"tokenizer_class": "NllbTokenizer",
|
30 |
+
"torch_dtype": "float32",
|
31 |
+
"transformers_version": "4.50.3",
|
32 |
+
"use_cache": true,
|
33 |
+
"vocab_size": 256206
|
34 |
+
}
|
model.safetensors
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:9ab97e5c781a6c686191cdee650641c1e77d9d920e81dd9446a5cec6b37606e7
|
3 |
+
size 2460354912
|
runs/Apr12_11-34-18_495dcc39ad78/events.out.tfevents.1744457676.495dcc39ad78.672.0
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:715dbfdfc024f9ae79f76826ef348af2450a0e249d6e5c10aca5c5430d106de3
|
3 |
+
size 4184
|
runs/Apr12_11-35-38_495dcc39ad78/events.out.tfevents.1744457741.495dcc39ad78.672.1
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:1538b717ae5617bc7937e7a3285a4ec480b0b900975e875785a44b7bfb29b5e2
|
3 |
+
size 4184
|
runs/Apr12_11-38-15_495dcc39ad78/events.out.tfevents.1744457905.495dcc39ad78.3074.0
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:90ca3140d5e701560a3005c8361488a3d5f32ef5958e06651207fdfc780d23fd
|
3 |
+
size 4184
|
runs/Apr12_11-39-52_495dcc39ad78/events.out.tfevents.1744458014.495dcc39ad78.3873.0
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:bd98ef4086fd10dc8f39f415083772244563581a073d7c0fb456ef316040fd60
|
3 |
+
size 6763
|
training_args.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d0e534f6f5640c0fab08af8ea74d2192d533e4ea93977fee46f4758b7ebdec54
|
3 |
+
size 5496
|