Vui Seng Chua
commited on
Commit
•
270b51c
1
Parent(s):
f9a5f98
add content
Browse files- README.md +97 -0
- all_results.json +11 -0
- config.json +27 -0
- eval_predictions.json +0 -0
- eval_results.json +6 -0
- merges.txt +0 -0
- pytorch_model.bin +3 -0
- special_tokens_map.json +1 -0
- tokenizer.json +0 -0
- tokenizer_config.json +1 -0
- train_results.json +8 -0
- trainer_state.json +0 -0
- training_args.bin +3 -0
- vocab.json +0 -0
README.md
ADDED
@@ -0,0 +1,97 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
license: mit
|
3 |
+
tags:
|
4 |
+
- generated_from_trainer
|
5 |
+
datasets:
|
6 |
+
- squad
|
7 |
+
model-index:
|
8 |
+
- name: run05-roberta-large-squadv1.1-sl384-ds128-e2-tbs16
|
9 |
+
results: []
|
10 |
+
---
|
11 |
+
|
12 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
13 |
+
should probably proofread and complete it, then remove this comment. -->
|
14 |
+
|
15 |
+
# run05-roberta-large-squadv1.1-sl384-ds128-e2-tbs16
|
16 |
+
|
17 |
+
This model is a fine-tuned version of [roberta-large](https://huggingface.co/roberta-large) on the squad dataset.
|
18 |
+
|
19 |
+
## Model description
|
20 |
+
|
21 |
+
More information needed
|
22 |
+
|
23 |
+
## Intended uses & limitations
|
24 |
+
|
25 |
+
More information needed
|
26 |
+
|
27 |
+
## Training and evaluation data
|
28 |
+
|
29 |
+
More information needed
|
30 |
+
|
31 |
+
## Training procedure
|
32 |
+
|
33 |
+
### Training hyperparameters
|
34 |
+
|
35 |
+
The following hyperparameters were used during training:
|
36 |
+
- learning_rate: 3e-05
|
37 |
+
- train_batch_size: 16
|
38 |
+
- eval_batch_size: 64
|
39 |
+
- seed: 42
|
40 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
41 |
+
- lr_scheduler_type: linear
|
42 |
+
- num_epochs: 2.0
|
43 |
+
- mixed_precision_training: Native AMP
|
44 |
+
|
45 |
+
### Training results
|
46 |
+
|
47 |
+
|
48 |
+
|
49 |
+
### Framework versions
|
50 |
+
|
51 |
+
- Transformers 4.18.0
|
52 |
+
- Pytorch 1.11.0+cu113
|
53 |
+
- Datasets 2.1.0
|
54 |
+
- Tokenizers 0.12.1
|
55 |
+
|
56 |
+
# Train
|
57 |
+
```bash
|
58 |
+
python run_qa.py \
|
59 |
+
--model_name_or_path roberta-large \
|
60 |
+
--dataset_name squad \
|
61 |
+
--do_eval \
|
62 |
+
--do_train \
|
63 |
+
--evaluation_strategy steps \
|
64 |
+
--eval_steps 500 \
|
65 |
+
--learning_rate 3e-5 \
|
66 |
+
--fp16 \
|
67 |
+
--num_train_epochs 2 \
|
68 |
+
--per_device_eval_batch_size 64 \
|
69 |
+
--per_device_train_batch_size 16 \
|
70 |
+
--max_seq_length 384 \
|
71 |
+
--doc_stride 128 \
|
72 |
+
--save_steps 1000 \
|
73 |
+
--logging_steps 1 \
|
74 |
+
--overwrite_output_dir \
|
75 |
+
--run_name $RUNID \
|
76 |
+
--output_dir $OUTDIR
|
77 |
+
```
|
78 |
+
|
79 |
+
# Eval
|
80 |
+
```bash
|
81 |
+
export CUDA_VISIBLE_DEVICES=0
|
82 |
+
|
83 |
+
MODEL=vuiseng9/roberta-l-squadv1.1
|
84 |
+
OUTDIR=eval-$(basename $MODEL)
|
85 |
+
WORKDIR=transformers/examples/pytorch/question-answering
|
86 |
+
cd $WORKDIR
|
87 |
+
|
88 |
+
nohup python run_qa.py \
|
89 |
+
--model_name_or_path $MODEL \
|
90 |
+
--dataset_name squad \
|
91 |
+
--do_eval \
|
92 |
+
--per_device_eval_batch_size 16 \
|
93 |
+
--max_seq_length 384 \
|
94 |
+
--doc_stride 128 \
|
95 |
+
--overwrite_output_dir \
|
96 |
+
--output_dir $OUTDIR 2>&1 | tee $OUTDIR/run.log &
|
97 |
+
```
|
all_results.json
ADDED
@@ -0,0 +1,11 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 2.0,
|
3 |
+
"eval_exact_match": 88.45789971617786,
|
4 |
+
"eval_f1": 94.30632459083219,
|
5 |
+
"eval_samples": 10790,
|
6 |
+
"train_loss": 0.7616665850008564,
|
7 |
+
"train_runtime": 4169.2181,
|
8 |
+
"train_samples": 88568,
|
9 |
+
"train_samples_per_second": 42.487,
|
10 |
+
"train_steps_per_second": 2.656
|
11 |
+
}
|
config.json
ADDED
@@ -0,0 +1,27 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "roberta-large",
|
3 |
+
"architectures": [
|
4 |
+
"RobertaForQuestionAnswering"
|
5 |
+
],
|
6 |
+
"attention_probs_dropout_prob": 0.1,
|
7 |
+
"bos_token_id": 0,
|
8 |
+
"classifier_dropout": null,
|
9 |
+
"eos_token_id": 2,
|
10 |
+
"hidden_act": "gelu",
|
11 |
+
"hidden_dropout_prob": 0.1,
|
12 |
+
"hidden_size": 1024,
|
13 |
+
"initializer_range": 0.02,
|
14 |
+
"intermediate_size": 4096,
|
15 |
+
"layer_norm_eps": 1e-05,
|
16 |
+
"max_position_embeddings": 514,
|
17 |
+
"model_type": "roberta",
|
18 |
+
"num_attention_heads": 16,
|
19 |
+
"num_hidden_layers": 24,
|
20 |
+
"pad_token_id": 1,
|
21 |
+
"position_embedding_type": "absolute",
|
22 |
+
"torch_dtype": "float32",
|
23 |
+
"transformers_version": "4.18.0",
|
24 |
+
"type_vocab_size": 1,
|
25 |
+
"use_cache": true,
|
26 |
+
"vocab_size": 50265
|
27 |
+
}
|
eval_predictions.json
ADDED
The diff for this file is too large to render.
See raw diff
|
|
eval_results.json
ADDED
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 2.0,
|
3 |
+
"eval_exact_match": 88.45789971617786,
|
4 |
+
"eval_f1": 94.30632459083219,
|
5 |
+
"eval_samples": 10790
|
6 |
+
}
|
merges.txt
ADDED
The diff for this file is too large to render.
See raw diff
|
|
pytorch_model.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:885be04aa71771d049d12c10423508ecdb4b6443415ea3443596812995bb826b
|
3 |
+
size 1417385329
|
special_tokens_map.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"bos_token": "<s>", "eos_token": "</s>", "unk_token": "<unk>", "sep_token": "</s>", "pad_token": "<pad>", "cls_token": "<s>", "mask_token": {"content": "<mask>", "single_word": false, "lstrip": true, "rstrip": false, "normalized": false}}
|
tokenizer.json
ADDED
The diff for this file is too large to render.
See raw diff
|
|
tokenizer_config.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"errors": "replace", "bos_token": "<s>", "eos_token": "</s>", "sep_token": "</s>", "cls_token": "<s>", "unk_token": "<unk>", "pad_token": "<pad>", "mask_token": "<mask>", "add_prefix_space": false, "trim_offsets": true, "model_max_length": 512, "special_tokens_map_file": null, "name_or_path": "roberta-large", "tokenizer_class": "RobertaTokenizer"}
|
train_results.json
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 2.0,
|
3 |
+
"train_loss": 0.7616665850008564,
|
4 |
+
"train_runtime": 4169.2181,
|
5 |
+
"train_samples": 88568,
|
6 |
+
"train_samples_per_second": 42.487,
|
7 |
+
"train_steps_per_second": 2.656
|
8 |
+
}
|
trainer_state.json
ADDED
The diff for this file is too large to render.
See raw diff
|
|
training_args.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:39147a6bac51bae479afda16858d1c769d5ca3bc514ba0589435908261d8c04a
|
3 |
+
size 3183
|
vocab.json
ADDED
The diff for this file is too large to render.
See raw diff
|
|