Upload 9 files
Browse files- UniPoll-t5/added_tokens.json +6 -0
- UniPoll-t5/config.json +32 -0
- UniPoll-t5/pytorch_model.bin +3 -0
- UniPoll-t5/special_tokens_map.json +7 -0
- UniPoll-t5/tokenizer_config.json +15 -0
- UniPoll-t5/trainer_state.json +1001 -0
- UniPoll-t5/training_args.bin +3 -0
- UniPoll-t5/vocab.txt +0 -0
- utils.py +64 -0
UniPoll-t5/added_tokens.json
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"<c>": 50001,
|
| 3 |
+
"<choices>": 50003,
|
| 4 |
+
"<title>": 50002,
|
| 5 |
+
"digit": 50000
|
| 6 |
+
}
|
UniPoll-t5/config.json
ADDED
|
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "/root/Documents/CODES/UniPoll/checkpoints/imxly/t5-pegasus",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"MT5ForConditionalGeneration"
|
| 5 |
+
],
|
| 6 |
+
"d_ff": 2048,
|
| 7 |
+
"d_kv": 64,
|
| 8 |
+
"d_model": 768,
|
| 9 |
+
"decoder_start_token_id": 0,
|
| 10 |
+
"dense_act_fn": "gelu_new",
|
| 11 |
+
"dropout_rate": 0.1,
|
| 12 |
+
"eos_token_id": 1,
|
| 13 |
+
"feed_forward_proj": "gated-gelu",
|
| 14 |
+
"initializer_factor": 1.0,
|
| 15 |
+
"is_encoder_decoder": true,
|
| 16 |
+
"is_gated_act": true,
|
| 17 |
+
"layer_norm_epsilon": 1e-06,
|
| 18 |
+
"model_type": "mt5",
|
| 19 |
+
"num_decoder_layers": 12,
|
| 20 |
+
"num_heads": 12,
|
| 21 |
+
"num_layers": 12,
|
| 22 |
+
"output_past": true,
|
| 23 |
+
"pad_token_id": 0,
|
| 24 |
+
"relative_attention_max_distance": 128,
|
| 25 |
+
"relative_attention_num_buckets": 32,
|
| 26 |
+
"tie_word_embeddings": false,
|
| 27 |
+
"tokenizer_class": "T5Tokenizer",
|
| 28 |
+
"torch_dtype": "float32",
|
| 29 |
+
"transformers_version": "4.22.1",
|
| 30 |
+
"use_cache": true,
|
| 31 |
+
"vocab_size": 50004
|
| 32 |
+
}
|
UniPoll-t5/pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1741c956f991a7ba239b38322657b7780961f9831914604a5adb75297f6f837e
|
| 3 |
+
size 1100236621
|
UniPoll-t5/special_tokens_map.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"cls_token": "[CLS]",
|
| 3 |
+
"mask_token": "[MASK]",
|
| 4 |
+
"pad_token": "[PAD]",
|
| 5 |
+
"sep_token": "[SEP]",
|
| 6 |
+
"unk_token": "[UNK]"
|
| 7 |
+
}
|
UniPoll-t5/tokenizer_config.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"cls_token": "[CLS]",
|
| 3 |
+
"do_basic_tokenize": true,
|
| 4 |
+
"do_lower_case": true,
|
| 5 |
+
"mask_token": "[MASK]",
|
| 6 |
+
"name_or_path": "/root/Documents/CODES/UniPoll/checkpoints/imxly/t5-pegasus",
|
| 7 |
+
"never_split": null,
|
| 8 |
+
"pad_token": "[PAD]",
|
| 9 |
+
"sep_token": "[SEP]",
|
| 10 |
+
"special_tokens_map_file": null,
|
| 11 |
+
"strip_accents": null,
|
| 12 |
+
"tokenize_chinese_chars": true,
|
| 13 |
+
"tokenizer_class": "T5PegasusTokenizer",
|
| 14 |
+
"unk_token": "[UNK]"
|
| 15 |
+
}
|
UniPoll-t5/trainer_state.json
ADDED
|
@@ -0,0 +1,1001 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 49.33,
|
| 3 |
+
"best_model_checkpoint": "outputs/UniPoll-t5/checkpoint-60760",
|
| 4 |
+
"epoch": 10.0,
|
| 5 |
+
"global_step": 60760,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.08,
|
| 12 |
+
"learning_rate": 2.9753127057274523e-05,
|
| 13 |
+
"loss": 16.5215,
|
| 14 |
+
"step": 500
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.16,
|
| 18 |
+
"learning_rate": 2.9506254114549046e-05,
|
| 19 |
+
"loss": 3.5706,
|
| 20 |
+
"step": 1000
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.25,
|
| 24 |
+
"learning_rate": 2.9259381171823572e-05,
|
| 25 |
+
"loss": 2.9107,
|
| 26 |
+
"step": 1500
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.33,
|
| 30 |
+
"learning_rate": 2.901250822909809e-05,
|
| 31 |
+
"loss": 2.6763,
|
| 32 |
+
"step": 2000
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.41,
|
| 36 |
+
"learning_rate": 2.8765635286372614e-05,
|
| 37 |
+
"loss": 2.4391,
|
| 38 |
+
"step": 2500
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.49,
|
| 42 |
+
"learning_rate": 2.8518762343647136e-05,
|
| 43 |
+
"loss": 2.2868,
|
| 44 |
+
"step": 3000
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.58,
|
| 48 |
+
"learning_rate": 2.827188940092166e-05,
|
| 49 |
+
"loss": 2.2185,
|
| 50 |
+
"step": 3500
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.66,
|
| 54 |
+
"learning_rate": 2.802501645819618e-05,
|
| 55 |
+
"loss": 2.0841,
|
| 56 |
+
"step": 4000
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.74,
|
| 60 |
+
"learning_rate": 2.7778143515470708e-05,
|
| 61 |
+
"loss": 2.013,
|
| 62 |
+
"step": 4500
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.82,
|
| 66 |
+
"learning_rate": 2.7531270572745227e-05,
|
| 67 |
+
"loss": 1.9524,
|
| 68 |
+
"step": 5000
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.91,
|
| 72 |
+
"learning_rate": 2.728439763001975e-05,
|
| 73 |
+
"loss": 1.9129,
|
| 74 |
+
"step": 5500
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.99,
|
| 78 |
+
"learning_rate": 2.7037524687294272e-05,
|
| 79 |
+
"loss": 1.8501,
|
| 80 |
+
"step": 6000
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 1.0,
|
| 84 |
+
"eval_choices_bleu1": 34.2743,
|
| 85 |
+
"eval_choices_bleu2": 27.071,
|
| 86 |
+
"eval_choices_bleu3": 23.8714,
|
| 87 |
+
"eval_choices_bleu4": 19.9582,
|
| 88 |
+
"eval_choices_rouge1": 43.056,
|
| 89 |
+
"eval_choices_rouge2": 30.797,
|
| 90 |
+
"eval_choices_rougeL": 40.79,
|
| 91 |
+
"eval_choices_rougeLsum": 30.362,
|
| 92 |
+
"eval_loss": 1.7218139171600342,
|
| 93 |
+
"eval_mean_rouge1": 44.783500000000004,
|
| 94 |
+
"eval_runtime": 613.1086,
|
| 95 |
+
"eval_samples_per_second": 3.303,
|
| 96 |
+
"eval_steps_per_second": 0.414,
|
| 97 |
+
"eval_title_bleu1": 39.1598,
|
| 98 |
+
"eval_title_bleu2": 24.8472,
|
| 99 |
+
"eval_title_bleu3": 17.1189,
|
| 100 |
+
"eval_title_bleu4": 11.4654,
|
| 101 |
+
"eval_title_rouge1": 46.511,
|
| 102 |
+
"eval_title_rouge2": 28.336,
|
| 103 |
+
"eval_title_rougeL": 43.814,
|
| 104 |
+
"eval_title_rougeLsum": 28.185,
|
| 105 |
+
"step": 6076
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 1.07,
|
| 109 |
+
"learning_rate": 2.6790651744568795e-05,
|
| 110 |
+
"loss": 1.7659,
|
| 111 |
+
"step": 6500
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 1.15,
|
| 115 |
+
"learning_rate": 2.654377880184332e-05,
|
| 116 |
+
"loss": 1.7064,
|
| 117 |
+
"step": 7000
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 1.23,
|
| 121 |
+
"learning_rate": 2.6296905859117843e-05,
|
| 122 |
+
"loss": 1.6739,
|
| 123 |
+
"step": 7500
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 1.32,
|
| 127 |
+
"learning_rate": 2.6050032916392366e-05,
|
| 128 |
+
"loss": 1.6589,
|
| 129 |
+
"step": 8000
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 1.4,
|
| 133 |
+
"learning_rate": 2.5803159973666885e-05,
|
| 134 |
+
"loss": 1.6193,
|
| 135 |
+
"step": 8500
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 1.48,
|
| 139 |
+
"learning_rate": 2.5556287030941408e-05,
|
| 140 |
+
"loss": 1.6175,
|
| 141 |
+
"step": 9000
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"epoch": 1.56,
|
| 145 |
+
"learning_rate": 2.530941408821593e-05,
|
| 146 |
+
"loss": 1.6042,
|
| 147 |
+
"step": 9500
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"epoch": 1.65,
|
| 151 |
+
"learning_rate": 2.5062541145490456e-05,
|
| 152 |
+
"loss": 1.5438,
|
| 153 |
+
"step": 10000
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 1.73,
|
| 157 |
+
"learning_rate": 2.481566820276498e-05,
|
| 158 |
+
"loss": 1.5275,
|
| 159 |
+
"step": 10500
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 1.81,
|
| 163 |
+
"learning_rate": 2.45687952600395e-05,
|
| 164 |
+
"loss": 1.5144,
|
| 165 |
+
"step": 11000
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 1.89,
|
| 169 |
+
"learning_rate": 2.432192231731402e-05,
|
| 170 |
+
"loss": 1.4911,
|
| 171 |
+
"step": 11500
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 1.97,
|
| 175 |
+
"learning_rate": 2.4075049374588544e-05,
|
| 176 |
+
"loss": 1.4301,
|
| 177 |
+
"step": 12000
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 2.0,
|
| 181 |
+
"eval_choices_bleu1": 36.2304,
|
| 182 |
+
"eval_choices_bleu2": 29.2436,
|
| 183 |
+
"eval_choices_bleu3": 26.0301,
|
| 184 |
+
"eval_choices_bleu4": 21.8164,
|
| 185 |
+
"eval_choices_rouge1": 45.164,
|
| 186 |
+
"eval_choices_rouge2": 33.104,
|
| 187 |
+
"eval_choices_rougeL": 42.835,
|
| 188 |
+
"eval_choices_rougeLsum": 32.498,
|
| 189 |
+
"eval_loss": 1.6042917966842651,
|
| 190 |
+
"eval_mean_rouge1": 47.2485,
|
| 191 |
+
"eval_runtime": 613.8133,
|
| 192 |
+
"eval_samples_per_second": 3.299,
|
| 193 |
+
"eval_steps_per_second": 0.414,
|
| 194 |
+
"eval_title_bleu1": 41.7754,
|
| 195 |
+
"eval_title_bleu2": 27.1102,
|
| 196 |
+
"eval_title_bleu3": 18.344,
|
| 197 |
+
"eval_title_bleu4": 12.3234,
|
| 198 |
+
"eval_title_rouge1": 49.333,
|
| 199 |
+
"eval_title_rouge2": 30.676,
|
| 200 |
+
"eval_title_rougeL": 46.561,
|
| 201 |
+
"eval_title_rougeLsum": 30.748,
|
| 202 |
+
"step": 12152
|
| 203 |
+
},
|
| 204 |
+
{
|
| 205 |
+
"epoch": 2.06,
|
| 206 |
+
"learning_rate": 2.3828176431863066e-05,
|
| 207 |
+
"loss": 1.4002,
|
| 208 |
+
"step": 12500
|
| 209 |
+
},
|
| 210 |
+
{
|
| 211 |
+
"epoch": 2.14,
|
| 212 |
+
"learning_rate": 2.3581303489137592e-05,
|
| 213 |
+
"loss": 1.3791,
|
| 214 |
+
"step": 13000
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"epoch": 2.22,
|
| 218 |
+
"learning_rate": 2.3334430546412115e-05,
|
| 219 |
+
"loss": 1.3332,
|
| 220 |
+
"step": 13500
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 2.3,
|
| 224 |
+
"learning_rate": 2.3087557603686637e-05,
|
| 225 |
+
"loss": 1.3577,
|
| 226 |
+
"step": 14000
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 2.39,
|
| 230 |
+
"learning_rate": 2.284068466096116e-05,
|
| 231 |
+
"loss": 1.3483,
|
| 232 |
+
"step": 14500
|
| 233 |
+
},
|
| 234 |
+
{
|
| 235 |
+
"epoch": 2.47,
|
| 236 |
+
"learning_rate": 2.259381171823568e-05,
|
| 237 |
+
"loss": 1.3584,
|
| 238 |
+
"step": 15000
|
| 239 |
+
},
|
| 240 |
+
{
|
| 241 |
+
"epoch": 2.55,
|
| 242 |
+
"learning_rate": 2.2346938775510205e-05,
|
| 243 |
+
"loss": 1.3186,
|
| 244 |
+
"step": 15500
|
| 245 |
+
},
|
| 246 |
+
{
|
| 247 |
+
"epoch": 2.63,
|
| 248 |
+
"learning_rate": 2.2100065832784728e-05,
|
| 249 |
+
"loss": 1.3283,
|
| 250 |
+
"step": 16000
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 2.72,
|
| 254 |
+
"learning_rate": 2.185319289005925e-05,
|
| 255 |
+
"loss": 1.2867,
|
| 256 |
+
"step": 16500
|
| 257 |
+
},
|
| 258 |
+
{
|
| 259 |
+
"epoch": 2.8,
|
| 260 |
+
"learning_rate": 2.1606319947333773e-05,
|
| 261 |
+
"loss": 1.2766,
|
| 262 |
+
"step": 17000
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 2.88,
|
| 266 |
+
"learning_rate": 2.1359447004608296e-05,
|
| 267 |
+
"loss": 1.3025,
|
| 268 |
+
"step": 17500
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 2.96,
|
| 272 |
+
"learning_rate": 2.111257406188282e-05,
|
| 273 |
+
"loss": 1.2736,
|
| 274 |
+
"step": 18000
|
| 275 |
+
},
|
| 276 |
+
{
|
| 277 |
+
"epoch": 3.0,
|
| 278 |
+
"eval_choices_bleu1": 36.9274,
|
| 279 |
+
"eval_choices_bleu2": 30.039,
|
| 280 |
+
"eval_choices_bleu3": 26.4107,
|
| 281 |
+
"eval_choices_bleu4": 22.0831,
|
| 282 |
+
"eval_choices_rouge1": 46.229,
|
| 283 |
+
"eval_choices_rouge2": 33.898,
|
| 284 |
+
"eval_choices_rougeL": 43.746,
|
| 285 |
+
"eval_choices_rougeLsum": 33.141,
|
| 286 |
+
"eval_loss": 1.5690504312515259,
|
| 287 |
+
"eval_mean_rouge1": 48.167500000000004,
|
| 288 |
+
"eval_runtime": 614.3575,
|
| 289 |
+
"eval_samples_per_second": 3.296,
|
| 290 |
+
"eval_steps_per_second": 0.413,
|
| 291 |
+
"eval_title_bleu1": 42.3694,
|
| 292 |
+
"eval_title_bleu2": 28.0839,
|
| 293 |
+
"eval_title_bleu3": 19.3916,
|
| 294 |
+
"eval_title_bleu4": 13.1408,
|
| 295 |
+
"eval_title_rouge1": 50.106,
|
| 296 |
+
"eval_title_rouge2": 31.608,
|
| 297 |
+
"eval_title_rougeL": 47.1,
|
| 298 |
+
"eval_title_rougeLsum": 31.399,
|
| 299 |
+
"step": 18228
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 3.04,
|
| 303 |
+
"learning_rate": 2.086570111915734e-05,
|
| 304 |
+
"loss": 1.2176,
|
| 305 |
+
"step": 18500
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 3.13,
|
| 309 |
+
"learning_rate": 2.0618828176431864e-05,
|
| 310 |
+
"loss": 1.2037,
|
| 311 |
+
"step": 19000
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 3.21,
|
| 315 |
+
"learning_rate": 2.0371955233706386e-05,
|
| 316 |
+
"loss": 1.2104,
|
| 317 |
+
"step": 19500
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 3.29,
|
| 321 |
+
"learning_rate": 2.012508229098091e-05,
|
| 322 |
+
"loss": 1.1522,
|
| 323 |
+
"step": 20000
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 3.37,
|
| 327 |
+
"learning_rate": 1.987820934825543e-05,
|
| 328 |
+
"loss": 1.1969,
|
| 329 |
+
"step": 20500
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 3.46,
|
| 333 |
+
"learning_rate": 1.9631336405529957e-05,
|
| 334 |
+
"loss": 1.1861,
|
| 335 |
+
"step": 21000
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 3.54,
|
| 339 |
+
"learning_rate": 1.9384463462804477e-05,
|
| 340 |
+
"loss": 1.1536,
|
| 341 |
+
"step": 21500
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 3.62,
|
| 345 |
+
"learning_rate": 1.9137590520079e-05,
|
| 346 |
+
"loss": 1.1394,
|
| 347 |
+
"step": 22000
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 3.7,
|
| 351 |
+
"learning_rate": 1.8890717577353522e-05,
|
| 352 |
+
"loss": 1.1229,
|
| 353 |
+
"step": 22500
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 3.79,
|
| 357 |
+
"learning_rate": 1.8643844634628044e-05,
|
| 358 |
+
"loss": 1.1233,
|
| 359 |
+
"step": 23000
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 3.87,
|
| 363 |
+
"learning_rate": 1.8396971691902567e-05,
|
| 364 |
+
"loss": 1.1265,
|
| 365 |
+
"step": 23500
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 3.95,
|
| 369 |
+
"learning_rate": 1.8150098749177093e-05,
|
| 370 |
+
"loss": 1.1438,
|
| 371 |
+
"step": 24000
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 4.0,
|
| 375 |
+
"eval_choices_bleu1": 38.2715,
|
| 376 |
+
"eval_choices_bleu2": 30.7803,
|
| 377 |
+
"eval_choices_bleu3": 26.9954,
|
| 378 |
+
"eval_choices_bleu4": 22.6626,
|
| 379 |
+
"eval_choices_rouge1": 46.767,
|
| 380 |
+
"eval_choices_rouge2": 34.301,
|
| 381 |
+
"eval_choices_rougeL": 44.062,
|
| 382 |
+
"eval_choices_rougeLsum": 33.605,
|
| 383 |
+
"eval_loss": 1.5627468824386597,
|
| 384 |
+
"eval_mean_rouge1": 48.6515,
|
| 385 |
+
"eval_runtime": 625.5501,
|
| 386 |
+
"eval_samples_per_second": 3.237,
|
| 387 |
+
"eval_steps_per_second": 0.406,
|
| 388 |
+
"eval_title_bleu1": 42.7056,
|
| 389 |
+
"eval_title_bleu2": 28.0993,
|
| 390 |
+
"eval_title_bleu3": 19.498,
|
| 391 |
+
"eval_title_bleu4": 12.7399,
|
| 392 |
+
"eval_title_rouge1": 50.536,
|
| 393 |
+
"eval_title_rouge2": 31.852,
|
| 394 |
+
"eval_title_rougeL": 47.453,
|
| 395 |
+
"eval_title_rougeLsum": 31.678,
|
| 396 |
+
"step": 24304
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 4.03,
|
| 400 |
+
"learning_rate": 1.7903225806451616e-05,
|
| 401 |
+
"loss": 1.0962,
|
| 402 |
+
"step": 24500
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 4.11,
|
| 406 |
+
"learning_rate": 1.7656352863726135e-05,
|
| 407 |
+
"loss": 1.0664,
|
| 408 |
+
"step": 25000
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 4.2,
|
| 412 |
+
"learning_rate": 1.7409479921000658e-05,
|
| 413 |
+
"loss": 1.0874,
|
| 414 |
+
"step": 25500
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 4.28,
|
| 418 |
+
"learning_rate": 1.716260697827518e-05,
|
| 419 |
+
"loss": 1.0608,
|
| 420 |
+
"step": 26000
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 4.36,
|
| 424 |
+
"learning_rate": 1.6915734035549703e-05,
|
| 425 |
+
"loss": 1.0367,
|
| 426 |
+
"step": 26500
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 4.44,
|
| 430 |
+
"learning_rate": 1.666886109282423e-05,
|
| 431 |
+
"loss": 1.0397,
|
| 432 |
+
"step": 27000
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"epoch": 4.53,
|
| 436 |
+
"learning_rate": 1.642198815009875e-05,
|
| 437 |
+
"loss": 1.0378,
|
| 438 |
+
"step": 27500
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"epoch": 4.61,
|
| 442 |
+
"learning_rate": 1.617511520737327e-05,
|
| 443 |
+
"loss": 1.0143,
|
| 444 |
+
"step": 28000
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 4.69,
|
| 448 |
+
"learning_rate": 1.5928242264647793e-05,
|
| 449 |
+
"loss": 1.0186,
|
| 450 |
+
"step": 28500
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 4.77,
|
| 454 |
+
"learning_rate": 1.5681369321922316e-05,
|
| 455 |
+
"loss": 1.0243,
|
| 456 |
+
"step": 29000
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 4.86,
|
| 460 |
+
"learning_rate": 1.5434496379196842e-05,
|
| 461 |
+
"loss": 1.038,
|
| 462 |
+
"step": 29500
|
| 463 |
+
},
|
| 464 |
+
{
|
| 465 |
+
"epoch": 4.94,
|
| 466 |
+
"learning_rate": 1.5187623436471364e-05,
|
| 467 |
+
"loss": 1.0026,
|
| 468 |
+
"step": 30000
|
| 469 |
+
},
|
| 470 |
+
{
|
| 471 |
+
"epoch": 5.0,
|
| 472 |
+
"eval_choices_bleu1": 37.513,
|
| 473 |
+
"eval_choices_bleu2": 30.5119,
|
| 474 |
+
"eval_choices_bleu3": 26.8236,
|
| 475 |
+
"eval_choices_bleu4": 22.3797,
|
| 476 |
+
"eval_choices_rouge1": 46.999,
|
| 477 |
+
"eval_choices_rouge2": 34.451,
|
| 478 |
+
"eval_choices_rougeL": 44.427,
|
| 479 |
+
"eval_choices_rougeLsum": 33.652,
|
| 480 |
+
"eval_loss": 1.5795127153396606,
|
| 481 |
+
"eval_mean_rouge1": 48.658500000000004,
|
| 482 |
+
"eval_runtime": 619.9047,
|
| 483 |
+
"eval_samples_per_second": 3.267,
|
| 484 |
+
"eval_steps_per_second": 0.41,
|
| 485 |
+
"eval_title_bleu1": 42.3561,
|
| 486 |
+
"eval_title_bleu2": 27.804,
|
| 487 |
+
"eval_title_bleu3": 19.0851,
|
| 488 |
+
"eval_title_bleu4": 12.6803,
|
| 489 |
+
"eval_title_rouge1": 50.318,
|
| 490 |
+
"eval_title_rouge2": 31.494,
|
| 491 |
+
"eval_title_rougeL": 47.297,
|
| 492 |
+
"eval_title_rougeLsum": 31.461,
|
| 493 |
+
"step": 30380
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 5.02,
|
| 497 |
+
"learning_rate": 1.4940750493745885e-05,
|
| 498 |
+
"loss": 1.0107,
|
| 499 |
+
"step": 30500
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 5.1,
|
| 503 |
+
"learning_rate": 1.4693877551020408e-05,
|
| 504 |
+
"loss": 0.9664,
|
| 505 |
+
"step": 31000
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 5.18,
|
| 509 |
+
"learning_rate": 1.4447004608294932e-05,
|
| 510 |
+
"loss": 0.9342,
|
| 511 |
+
"step": 31500
|
| 512 |
+
},
|
| 513 |
+
{
|
| 514 |
+
"epoch": 5.27,
|
| 515 |
+
"learning_rate": 1.4200131665569453e-05,
|
| 516 |
+
"loss": 0.9632,
|
| 517 |
+
"step": 32000
|
| 518 |
+
},
|
| 519 |
+
{
|
| 520 |
+
"epoch": 5.35,
|
| 521 |
+
"learning_rate": 1.3953258722843976e-05,
|
| 522 |
+
"loss": 0.9448,
|
| 523 |
+
"step": 32500
|
| 524 |
+
},
|
| 525 |
+
{
|
| 526 |
+
"epoch": 5.43,
|
| 527 |
+
"learning_rate": 1.37063857801185e-05,
|
| 528 |
+
"loss": 0.9669,
|
| 529 |
+
"step": 33000
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 5.51,
|
| 533 |
+
"learning_rate": 1.3459512837393023e-05,
|
| 534 |
+
"loss": 0.9361,
|
| 535 |
+
"step": 33500
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 5.6,
|
| 539 |
+
"learning_rate": 1.3212639894667544e-05,
|
| 540 |
+
"loss": 0.9445,
|
| 541 |
+
"step": 34000
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 5.68,
|
| 545 |
+
"learning_rate": 1.2965766951942068e-05,
|
| 546 |
+
"loss": 0.9515,
|
| 547 |
+
"step": 34500
|
| 548 |
+
},
|
| 549 |
+
{
|
| 550 |
+
"epoch": 5.76,
|
| 551 |
+
"learning_rate": 1.271889400921659e-05,
|
| 552 |
+
"loss": 0.9488,
|
| 553 |
+
"step": 35000
|
| 554 |
+
},
|
| 555 |
+
{
|
| 556 |
+
"epoch": 5.84,
|
| 557 |
+
"learning_rate": 1.2472021066491112e-05,
|
| 558 |
+
"loss": 0.9145,
|
| 559 |
+
"step": 35500
|
| 560 |
+
},
|
| 561 |
+
{
|
| 562 |
+
"epoch": 5.92,
|
| 563 |
+
"learning_rate": 1.2225148123765636e-05,
|
| 564 |
+
"loss": 0.9493,
|
| 565 |
+
"step": 36000
|
| 566 |
+
},
|
| 567 |
+
{
|
| 568 |
+
"epoch": 6.0,
|
| 569 |
+
"eval_choices_bleu1": 38.7421,
|
| 570 |
+
"eval_choices_bleu2": 31.1954,
|
| 571 |
+
"eval_choices_bleu3": 27.0903,
|
| 572 |
+
"eval_choices_bleu4": 22.4639,
|
| 573 |
+
"eval_choices_rouge1": 47.617,
|
| 574 |
+
"eval_choices_rouge2": 34.86,
|
| 575 |
+
"eval_choices_rougeL": 44.679,
|
| 576 |
+
"eval_choices_rougeLsum": 34.003,
|
| 577 |
+
"eval_loss": 1.5864702463150024,
|
| 578 |
+
"eval_mean_rouge1": 48.966499999999996,
|
| 579 |
+
"eval_runtime": 615.3469,
|
| 580 |
+
"eval_samples_per_second": 3.291,
|
| 581 |
+
"eval_steps_per_second": 0.413,
|
| 582 |
+
"eval_title_bleu1": 42.4031,
|
| 583 |
+
"eval_title_bleu2": 28.2011,
|
| 584 |
+
"eval_title_bleu3": 19.3663,
|
| 585 |
+
"eval_title_bleu4": 12.8818,
|
| 586 |
+
"eval_title_rouge1": 50.316,
|
| 587 |
+
"eval_title_rouge2": 31.841,
|
| 588 |
+
"eval_title_rougeL": 47.179,
|
| 589 |
+
"eval_title_rougeLsum": 31.509,
|
| 590 |
+
"step": 36456
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 6.01,
|
| 594 |
+
"learning_rate": 1.1978275181040158e-05,
|
| 595 |
+
"loss": 0.9359,
|
| 596 |
+
"step": 36500
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 6.09,
|
| 600 |
+
"learning_rate": 1.1731402238314681e-05,
|
| 601 |
+
"loss": 0.9096,
|
| 602 |
+
"step": 37000
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 6.17,
|
| 606 |
+
"learning_rate": 1.1484529295589204e-05,
|
| 607 |
+
"loss": 0.9069,
|
| 608 |
+
"step": 37500
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 6.25,
|
| 612 |
+
"learning_rate": 1.1237656352863726e-05,
|
| 613 |
+
"loss": 0.89,
|
| 614 |
+
"step": 38000
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 6.34,
|
| 618 |
+
"learning_rate": 1.099078341013825e-05,
|
| 619 |
+
"loss": 0.8761,
|
| 620 |
+
"step": 38500
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 6.42,
|
| 624 |
+
"learning_rate": 1.0743910467412772e-05,
|
| 625 |
+
"loss": 0.8919,
|
| 626 |
+
"step": 39000
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 6.5,
|
| 630 |
+
"learning_rate": 1.0497037524687294e-05,
|
| 631 |
+
"loss": 0.8869,
|
| 632 |
+
"step": 39500
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 6.58,
|
| 636 |
+
"learning_rate": 1.0250164581961818e-05,
|
| 637 |
+
"loss": 0.8912,
|
| 638 |
+
"step": 40000
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 6.67,
|
| 642 |
+
"learning_rate": 1.000329163923634e-05,
|
| 643 |
+
"loss": 0.8885,
|
| 644 |
+
"step": 40500
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 6.75,
|
| 648 |
+
"learning_rate": 9.756418696510862e-06,
|
| 649 |
+
"loss": 0.8691,
|
| 650 |
+
"step": 41000
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 6.83,
|
| 654 |
+
"learning_rate": 9.509545753785386e-06,
|
| 655 |
+
"loss": 0.8882,
|
| 656 |
+
"step": 41500
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"epoch": 6.91,
|
| 660 |
+
"learning_rate": 9.262672811059907e-06,
|
| 661 |
+
"loss": 0.8822,
|
| 662 |
+
"step": 42000
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 6.99,
|
| 666 |
+
"learning_rate": 9.01579986833443e-06,
|
| 667 |
+
"loss": 0.8622,
|
| 668 |
+
"step": 42500
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 7.0,
|
| 672 |
+
"eval_choices_bleu1": 38.5191,
|
| 673 |
+
"eval_choices_bleu2": 31.1743,
|
| 674 |
+
"eval_choices_bleu3": 27.2188,
|
| 675 |
+
"eval_choices_bleu4": 22.5257,
|
| 676 |
+
"eval_choices_rouge1": 47.345,
|
| 677 |
+
"eval_choices_rouge2": 34.733,
|
| 678 |
+
"eval_choices_rougeL": 44.637,
|
| 679 |
+
"eval_choices_rougeLsum": 33.956,
|
| 680 |
+
"eval_loss": 1.6173765659332275,
|
| 681 |
+
"eval_mean_rouge1": 49.0085,
|
| 682 |
+
"eval_runtime": 624.2904,
|
| 683 |
+
"eval_samples_per_second": 3.244,
|
| 684 |
+
"eval_steps_per_second": 0.407,
|
| 685 |
+
"eval_title_bleu1": 42.7395,
|
| 686 |
+
"eval_title_bleu2": 28.2813,
|
| 687 |
+
"eval_title_bleu3": 19.4836,
|
| 688 |
+
"eval_title_bleu4": 12.7587,
|
| 689 |
+
"eval_title_rouge1": 50.672,
|
| 690 |
+
"eval_title_rouge2": 32.013,
|
| 691 |
+
"eval_title_rougeL": 47.817,
|
| 692 |
+
"eval_title_rougeLsum": 31.899,
|
| 693 |
+
"step": 42532
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 7.08,
|
| 697 |
+
"learning_rate": 8.768926925608954e-06,
|
| 698 |
+
"loss": 0.8468,
|
| 699 |
+
"step": 43000
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 7.16,
|
| 703 |
+
"learning_rate": 8.522053982883475e-06,
|
| 704 |
+
"loss": 0.8209,
|
| 705 |
+
"step": 43500
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 7.24,
|
| 709 |
+
"learning_rate": 8.275181040158e-06,
|
| 710 |
+
"loss": 0.8379,
|
| 711 |
+
"step": 44000
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 7.32,
|
| 715 |
+
"learning_rate": 8.028308097432522e-06,
|
| 716 |
+
"loss": 0.8544,
|
| 717 |
+
"step": 44500
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 7.41,
|
| 721 |
+
"learning_rate": 7.781435154707045e-06,
|
| 722 |
+
"loss": 0.8394,
|
| 723 |
+
"step": 45000
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 7.49,
|
| 727 |
+
"learning_rate": 7.534562211981568e-06,
|
| 728 |
+
"loss": 0.8285,
|
| 729 |
+
"step": 45500
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 7.57,
|
| 733 |
+
"learning_rate": 7.28768926925609e-06,
|
| 734 |
+
"loss": 0.8333,
|
| 735 |
+
"step": 46000
|
| 736 |
+
},
|
| 737 |
+
{
|
| 738 |
+
"epoch": 7.65,
|
| 739 |
+
"learning_rate": 7.0408163265306125e-06,
|
| 740 |
+
"loss": 0.8359,
|
| 741 |
+
"step": 46500
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 7.74,
|
| 745 |
+
"learning_rate": 6.793943383805135e-06,
|
| 746 |
+
"loss": 0.8176,
|
| 747 |
+
"step": 47000
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 7.82,
|
| 751 |
+
"learning_rate": 6.547070441079658e-06,
|
| 752 |
+
"loss": 0.842,
|
| 753 |
+
"step": 47500
|
| 754 |
+
},
|
| 755 |
+
{
|
| 756 |
+
"epoch": 7.9,
|
| 757 |
+
"learning_rate": 6.30019749835418e-06,
|
| 758 |
+
"loss": 0.8351,
|
| 759 |
+
"step": 48000
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 7.98,
|
| 763 |
+
"learning_rate": 6.053324555628703e-06,
|
| 764 |
+
"loss": 0.8367,
|
| 765 |
+
"step": 48500
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 8.0,
|
| 769 |
+
"eval_choices_bleu1": 38.7622,
|
| 770 |
+
"eval_choices_bleu2": 31.0867,
|
| 771 |
+
"eval_choices_bleu3": 26.9654,
|
| 772 |
+
"eval_choices_bleu4": 22.4556,
|
| 773 |
+
"eval_choices_rouge1": 47.531,
|
| 774 |
+
"eval_choices_rouge2": 34.756,
|
| 775 |
+
"eval_choices_rougeL": 44.649,
|
| 776 |
+
"eval_choices_rougeLsum": 33.959,
|
| 777 |
+
"eval_loss": 1.6293848752975464,
|
| 778 |
+
"eval_mean_rouge1": 49.2605,
|
| 779 |
+
"eval_runtime": 622.1469,
|
| 780 |
+
"eval_samples_per_second": 3.255,
|
| 781 |
+
"eval_steps_per_second": 0.408,
|
| 782 |
+
"eval_title_bleu1": 43.0901,
|
| 783 |
+
"eval_title_bleu2": 28.6471,
|
| 784 |
+
"eval_title_bleu3": 19.7616,
|
| 785 |
+
"eval_title_bleu4": 12.9726,
|
| 786 |
+
"eval_title_rouge1": 50.99,
|
| 787 |
+
"eval_title_rouge2": 32.44,
|
| 788 |
+
"eval_title_rougeL": 47.948,
|
| 789 |
+
"eval_title_rougeLsum": 32.24,
|
| 790 |
+
"step": 48608
|
| 791 |
+
},
|
| 792 |
+
{
|
| 793 |
+
"epoch": 8.06,
|
| 794 |
+
"learning_rate": 5.8064516129032256e-06,
|
| 795 |
+
"loss": 0.8004,
|
| 796 |
+
"step": 49000
|
| 797 |
+
},
|
| 798 |
+
{
|
| 799 |
+
"epoch": 8.15,
|
| 800 |
+
"learning_rate": 5.559578670177749e-06,
|
| 801 |
+
"loss": 0.789,
|
| 802 |
+
"step": 49500
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 8.23,
|
| 806 |
+
"learning_rate": 5.312705727452272e-06,
|
| 807 |
+
"loss": 0.7843,
|
| 808 |
+
"step": 50000
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 8.31,
|
| 812 |
+
"learning_rate": 5.065832784726793e-06,
|
| 813 |
+
"loss": 0.8102,
|
| 814 |
+
"step": 50500
|
| 815 |
+
},
|
| 816 |
+
{
|
| 817 |
+
"epoch": 8.39,
|
| 818 |
+
"learning_rate": 4.818959842001317e-06,
|
| 819 |
+
"loss": 0.7979,
|
| 820 |
+
"step": 51000
|
| 821 |
+
},
|
| 822 |
+
{
|
| 823 |
+
"epoch": 8.48,
|
| 824 |
+
"learning_rate": 4.5720868992758395e-06,
|
| 825 |
+
"loss": 0.8262,
|
| 826 |
+
"step": 51500
|
| 827 |
+
},
|
| 828 |
+
{
|
| 829 |
+
"epoch": 8.56,
|
| 830 |
+
"learning_rate": 4.325213956550362e-06,
|
| 831 |
+
"loss": 0.8149,
|
| 832 |
+
"step": 52000
|
| 833 |
+
},
|
| 834 |
+
{
|
| 835 |
+
"epoch": 8.64,
|
| 836 |
+
"learning_rate": 4.078341013824885e-06,
|
| 837 |
+
"loss": 0.8178,
|
| 838 |
+
"step": 52500
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 8.72,
|
| 842 |
+
"learning_rate": 3.831468071099407e-06,
|
| 843 |
+
"loss": 0.7796,
|
| 844 |
+
"step": 53000
|
| 845 |
+
},
|
| 846 |
+
{
|
| 847 |
+
"epoch": 8.81,
|
| 848 |
+
"learning_rate": 3.5845951283739304e-06,
|
| 849 |
+
"loss": 0.8036,
|
| 850 |
+
"step": 53500
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"epoch": 8.89,
|
| 854 |
+
"learning_rate": 3.337722185648453e-06,
|
| 855 |
+
"loss": 0.8169,
|
| 856 |
+
"step": 54000
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 8.97,
|
| 860 |
+
"learning_rate": 3.0908492429229756e-06,
|
| 861 |
+
"loss": 0.8138,
|
| 862 |
+
"step": 54500
|
| 863 |
+
},
|
| 864 |
+
{
|
| 865 |
+
"epoch": 9.0,
|
| 866 |
+
"eval_choices_bleu1": 38.8037,
|
| 867 |
+
"eval_choices_bleu2": 31.1809,
|
| 868 |
+
"eval_choices_bleu3": 27.0585,
|
| 869 |
+
"eval_choices_bleu4": 22.3774,
|
| 870 |
+
"eval_choices_rouge1": 47.629,
|
| 871 |
+
"eval_choices_rouge2": 34.778,
|
| 872 |
+
"eval_choices_rougeL": 44.722,
|
| 873 |
+
"eval_choices_rougeLsum": 33.944,
|
| 874 |
+
"eval_loss": 1.6443856954574585,
|
| 875 |
+
"eval_mean_rouge1": 49.32,
|
| 876 |
+
"eval_runtime": 624.8802,
|
| 877 |
+
"eval_samples_per_second": 3.241,
|
| 878 |
+
"eval_steps_per_second": 0.406,
|
| 879 |
+
"eval_title_bleu1": 43.0453,
|
| 880 |
+
"eval_title_bleu2": 28.6763,
|
| 881 |
+
"eval_title_bleu3": 19.7491,
|
| 882 |
+
"eval_title_bleu4": 13.004,
|
| 883 |
+
"eval_title_rouge1": 51.011,
|
| 884 |
+
"eval_title_rouge2": 32.398,
|
| 885 |
+
"eval_title_rougeL": 47.974,
|
| 886 |
+
"eval_title_rougeLsum": 32.251,
|
| 887 |
+
"step": 54684
|
| 888 |
+
},
|
| 889 |
+
{
|
| 890 |
+
"epoch": 9.05,
|
| 891 |
+
"learning_rate": 2.8439763001974987e-06,
|
| 892 |
+
"loss": 0.7764,
|
| 893 |
+
"step": 55000
|
| 894 |
+
},
|
| 895 |
+
{
|
| 896 |
+
"epoch": 9.13,
|
| 897 |
+
"learning_rate": 2.5971033574720213e-06,
|
| 898 |
+
"loss": 0.7939,
|
| 899 |
+
"step": 55500
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 9.22,
|
| 903 |
+
"learning_rate": 2.3502304147465435e-06,
|
| 904 |
+
"loss": 0.8069,
|
| 905 |
+
"step": 56000
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 9.3,
|
| 909 |
+
"learning_rate": 2.1033574720210665e-06,
|
| 910 |
+
"loss": 0.7649,
|
| 911 |
+
"step": 56500
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 9.38,
|
| 915 |
+
"learning_rate": 1.8564845292955893e-06,
|
| 916 |
+
"loss": 0.7977,
|
| 917 |
+
"step": 57000
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 9.46,
|
| 921 |
+
"learning_rate": 1.609611586570112e-06,
|
| 922 |
+
"loss": 0.803,
|
| 923 |
+
"step": 57500
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 9.55,
|
| 927 |
+
"learning_rate": 1.3627386438446346e-06,
|
| 928 |
+
"loss": 0.7813,
|
| 929 |
+
"step": 58000
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 9.63,
|
| 933 |
+
"learning_rate": 1.1158657011191574e-06,
|
| 934 |
+
"loss": 0.7645,
|
| 935 |
+
"step": 58500
|
| 936 |
+
},
|
| 937 |
+
{
|
| 938 |
+
"epoch": 9.71,
|
| 939 |
+
"learning_rate": 8.689927583936801e-07,
|
| 940 |
+
"loss": 0.7865,
|
| 941 |
+
"step": 59000
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 9.79,
|
| 945 |
+
"learning_rate": 6.221198156682028e-07,
|
| 946 |
+
"loss": 0.7818,
|
| 947 |
+
"step": 59500
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 9.87,
|
| 951 |
+
"learning_rate": 3.752468729427255e-07,
|
| 952 |
+
"loss": 0.7572,
|
| 953 |
+
"step": 60000
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 9.96,
|
| 957 |
+
"learning_rate": 1.283739302172482e-07,
|
| 958 |
+
"loss": 0.7788,
|
| 959 |
+
"step": 60500
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 10.0,
|
| 963 |
+
"eval_choices_bleu1": 38.8515,
|
| 964 |
+
"eval_choices_bleu2": 31.2992,
|
| 965 |
+
"eval_choices_bleu3": 27.2073,
|
| 966 |
+
"eval_choices_bleu4": 22.4633,
|
| 967 |
+
"eval_choices_rouge1": 47.688,
|
| 968 |
+
"eval_choices_rouge2": 34.913,
|
| 969 |
+
"eval_choices_rougeL": 44.77,
|
| 970 |
+
"eval_choices_rougeLsum": 34.038,
|
| 971 |
+
"eval_loss": 1.6531261205673218,
|
| 972 |
+
"eval_mean_rouge1": 49.33,
|
| 973 |
+
"eval_runtime": 620.225,
|
| 974 |
+
"eval_samples_per_second": 3.265,
|
| 975 |
+
"eval_steps_per_second": 0.41,
|
| 976 |
+
"eval_title_bleu1": 43.0602,
|
| 977 |
+
"eval_title_bleu2": 28.6495,
|
| 978 |
+
"eval_title_bleu3": 19.8146,
|
| 979 |
+
"eval_title_bleu4": 13.144,
|
| 980 |
+
"eval_title_rouge1": 50.972,
|
| 981 |
+
"eval_title_rouge2": 32.375,
|
| 982 |
+
"eval_title_rougeL": 47.919,
|
| 983 |
+
"eval_title_rougeLsum": 32.201,
|
| 984 |
+
"step": 60760
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 10.0,
|
| 988 |
+
"step": 60760,
|
| 989 |
+
"total_flos": 1.262557342857001e+17,
|
| 990 |
+
"train_loss": 1.288209796626761,
|
| 991 |
+
"train_runtime": 17551.2786,
|
| 992 |
+
"train_samples_per_second": 27.692,
|
| 993 |
+
"train_steps_per_second": 3.462
|
| 994 |
+
}
|
| 995 |
+
],
|
| 996 |
+
"max_steps": 60760,
|
| 997 |
+
"num_train_epochs": 10,
|
| 998 |
+
"total_flos": 1.262557342857001e+17,
|
| 999 |
+
"trial_name": null,
|
| 1000 |
+
"trial_params": null
|
| 1001 |
+
}
|
UniPoll-t5/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:53d331e9447f18c1699297a18369f397bdf5607c47ff32a23a2d6d9c4d795179
|
| 3 |
+
size 3503
|
UniPoll-t5/vocab.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
utils.py
ADDED
|
@@ -0,0 +1,64 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import os
|
| 2 |
+
import json
|
| 3 |
+
from pathlib import Path
|
| 4 |
+
from loguru import logger
|
| 5 |
+
from typing import Callable, Iterable, List
|
| 6 |
+
|
| 7 |
+
|
| 8 |
+
def lmap(f: Callable, x: Iterable) -> List:
|
| 9 |
+
"""list(map(f, x))"""
|
| 10 |
+
return list(map(f, x))
|
| 11 |
+
|
| 12 |
+
def write_txt_file(ordered_tgt, path):
|
| 13 |
+
f = Path(path).open("w")
|
| 14 |
+
for ln in ordered_tgt:
|
| 15 |
+
f.write(ln + "\n")
|
| 16 |
+
f.flush()
|
| 17 |
+
|
| 18 |
+
|
| 19 |
+
def save_json(content, path, indent=4, **json_dump_kwargs):
|
| 20 |
+
with open(path, "w") as f:
|
| 21 |
+
json.dump(content, f, indent=indent, sort_keys=True, **json_dump_kwargs)
|
| 22 |
+
|
| 23 |
+
|
| 24 |
+
def handle_metrics(split, metrics, output_dir):
|
| 25 |
+
"""
|
| 26 |
+
Log and save metrics
|
| 27 |
+
|
| 28 |
+
Args:
|
| 29 |
+
- split: one of train, val, test
|
| 30 |
+
- metrics: metrics dict
|
| 31 |
+
- output_dir: where to save the metrics
|
| 32 |
+
"""
|
| 33 |
+
|
| 34 |
+
logger.info(f"***** {split} metrics *****")
|
| 35 |
+
for key in sorted(metrics.keys()):
|
| 36 |
+
logger.info(f" {key} = {metrics[key]}")
|
| 37 |
+
save_json(metrics, os.path.join(output_dir, f"results_{split}.json"))
|
| 38 |
+
|
| 39 |
+
|
| 40 |
+
import shutil
|
| 41 |
+
def delete_checkpoints(model_dir):
|
| 42 |
+
checkpoints = [folder for folder in os.listdir(model_dir) if folder.split("-")[0]=="checkpoint"]
|
| 43 |
+
logger.info(f"Deleting checkpoints.\n{checkpoints}")
|
| 44 |
+
for checkpoint in checkpoints:
|
| 45 |
+
shutil.rmtree(os.path.join(model_dir, checkpoint))
|
| 46 |
+
|
| 47 |
+
|
| 48 |
+
import jieba
|
| 49 |
+
from functools import partial
|
| 50 |
+
from transformers import BertTokenizer
|
| 51 |
+
|
| 52 |
+
class T5PegasusTokenizer(BertTokenizer):
|
| 53 |
+
def __init__(self, *args, **kwargs):
|
| 54 |
+
super().__init__(*args, **kwargs)
|
| 55 |
+
self.pre_tokenizer = partial(jieba.cut, HMM=False)
|
| 56 |
+
|
| 57 |
+
def _tokenize(self, text, *arg, **kwargs):
|
| 58 |
+
split_tokens = []
|
| 59 |
+
for text in self.pre_tokenizer(text):
|
| 60 |
+
if text in self.vocab:
|
| 61 |
+
split_tokens.append(text)
|
| 62 |
+
else:
|
| 63 |
+
split_tokens.extend(super()._tokenize(text))
|
| 64 |
+
return split_tokens
|