Upload folder using huggingface_hub
Browse files- .gitattributes +1 -0
- .ipynb_checkpoints/conf-checkpoint.yml +80 -0
- 2025-01_conversations_2024_2.jsonl +3 -0
- conf.yml +80 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00000-of-00044.arrow +3 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00001-of-00044.arrow +3 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00002-of-00044.arrow +3 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00003-of-00044.arrow +3 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00004-of-00044.arrow +3 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00005-of-00044.arrow +3 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00006-of-00044.arrow +3 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00007-of-00044.arrow +3 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00008-of-00044.arrow +3 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00009-of-00044.arrow +3 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00010-of-00044.arrow +3 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00011-of-00044.arrow +3 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00012-of-00044.arrow +3 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00013-of-00044.arrow +3 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00014-of-00044.arrow +3 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00015-of-00044.arrow +3 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00016-of-00044.arrow +3 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00017-of-00044.arrow +3 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00018-of-00044.arrow +3 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00019-of-00044.arrow +3 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00020-of-00044.arrow +3 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00021-of-00044.arrow +3 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00022-of-00044.arrow +3 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00023-of-00044.arrow +3 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00024-of-00044.arrow +3 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00025-of-00044.arrow +3 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00026-of-00044.arrow +3 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00027-of-00044.arrow +3 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00028-of-00044.arrow +3 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00029-of-00044.arrow +3 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00030-of-00044.arrow +3 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00031-of-00044.arrow +3 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00032-of-00044.arrow +3 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00033-of-00044.arrow +3 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00034-of-00044.arrow +3 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00035-of-00044.arrow +3 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00036-of-00044.arrow +3 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00037-of-00044.arrow +3 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00038-of-00044.arrow +3 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00039-of-00044.arrow +3 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00040-of-00044.arrow +3 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00041-of-00044.arrow +3 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00042-of-00044.arrow +3 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00043-of-00044.arrow +3 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/dataset_info.json +40 -0
- last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/state.json +142 -0
.gitattributes
CHANGED
@@ -38,3 +38,4 @@ checkpoint-2727/tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
|
38 |
checkpoint-3636/tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
39 |
checkpoint-909/tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
40 |
tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
|
|
|
38 |
checkpoint-3636/tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
39 |
checkpoint-909/tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
40 |
tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
41 |
+
2025-01_conversations_2024_2.jsonl filter=lfs diff=lfs merge=lfs -text
|
.ipynb_checkpoints/conf-checkpoint.yml
ADDED
@@ -0,0 +1,80 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
base_model: ./meta-llama_Llama-3.1-8B
|
2 |
+
# optionally might have model_type or tokenizer_type
|
3 |
+
model_type: LlamaForCausalLM
|
4 |
+
tokenizer_type: AutoTokenizer
|
5 |
+
# Automatically upload checkpoint and final model to HF
|
6 |
+
# hub_model_id: username/custom_model_name
|
7 |
+
|
8 |
+
load_in_8bit: false
|
9 |
+
load_in_4bit: false
|
10 |
+
strict: false
|
11 |
+
|
12 |
+
datasets:
|
13 |
+
- path: 2025-01_conversations_2024_2.jsonl
|
14 |
+
type: chat_template
|
15 |
+
chat_template: tokenizer_default
|
16 |
+
field_messages: conversations
|
17 |
+
message_field_role: from
|
18 |
+
message_field_content: value
|
19 |
+
roles:
|
20 |
+
<|autheur|>:
|
21 |
+
- human
|
22 |
+
<|khey|>:
|
23 |
+
- gpt
|
24 |
+
<|sujet|>:
|
25 |
+
- system
|
26 |
+
dataset_prepared_path:
|
27 |
+
val_set_size: 0.05
|
28 |
+
output_dir: ./outputs/
|
29 |
+
dataset_prepared_path: last_run_prepared
|
30 |
+
|
31 |
+
sequence_len: 8192
|
32 |
+
sample_packing: true
|
33 |
+
pad_to_sequence_len: true
|
34 |
+
|
35 |
+
wandb_project: JVCGPT Medium 8b v2
|
36 |
+
wandb_entity:
|
37 |
+
wandb_watch:
|
38 |
+
wandb_name:
|
39 |
+
wandb_log_model:
|
40 |
+
|
41 |
+
gradient_accumulation_steps: 1
|
42 |
+
micro_batch_size: 1
|
43 |
+
num_epochs: 2
|
44 |
+
optimizer: paged_adamw_8bit
|
45 |
+
lr_scheduler: cosine
|
46 |
+
learning_rate: 1e-5
|
47 |
+
|
48 |
+
train_on_inputs: true
|
49 |
+
group_by_length: false
|
50 |
+
bf16: auto
|
51 |
+
fp16:
|
52 |
+
tf32: false
|
53 |
+
|
54 |
+
gradient_checkpointing: unsloth
|
55 |
+
gradient_checkpointing_kwargs:
|
56 |
+
use_reentrant: false
|
57 |
+
early_stopping_patience:
|
58 |
+
resume_from_checkpoint:
|
59 |
+
local_rank:
|
60 |
+
logging_steps: 1
|
61 |
+
xformers_attention:
|
62 |
+
flash_attention: true
|
63 |
+
s2_attention:
|
64 |
+
|
65 |
+
warmup_steps: 100
|
66 |
+
eval_table_size:
|
67 |
+
saves_per_epoch: 20
|
68 |
+
debug:
|
69 |
+
deepspeed:
|
70 |
+
weight_decay: 0.0
|
71 |
+
fsdp:
|
72 |
+
fsdp_config:
|
73 |
+
special_tokens:
|
74 |
+
pad_token: <|end_of_text|>
|
75 |
+
save_safetensors: true
|
76 |
+
save_total_limit: 5
|
77 |
+
resume_from_checkpoint: ./outputs/checkpoint-9999
|
78 |
+
# If resume_from_checkpoint isn't set and you simply want it to start where it left off.
|
79 |
+
# Be careful with this being turned on between different models.
|
80 |
+
auto_resume_from_checkpoints: true
|
2025-01_conversations_2024_2.jsonl
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d0c8af3b6b99bf9478f9f404cc42d257695ee679797f6bb7f4ff3e3bf34b6063
|
3 |
+
size 8576158452
|
conf.yml
ADDED
@@ -0,0 +1,80 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
base_model: ./meta-llama_Llama-3.1-8B
|
2 |
+
# optionally might have model_type or tokenizer_type
|
3 |
+
model_type: LlamaForCausalLM
|
4 |
+
tokenizer_type: AutoTokenizer
|
5 |
+
# Automatically upload checkpoint and final model to HF
|
6 |
+
# hub_model_id: username/custom_model_name
|
7 |
+
|
8 |
+
load_in_8bit: false
|
9 |
+
load_in_4bit: false
|
10 |
+
strict: false
|
11 |
+
|
12 |
+
datasets:
|
13 |
+
- path: 2025-01_conversations_2024_2.jsonl
|
14 |
+
type: chat_template
|
15 |
+
chat_template: tokenizer_default
|
16 |
+
field_messages: conversations
|
17 |
+
message_field_role: from
|
18 |
+
message_field_content: value
|
19 |
+
roles:
|
20 |
+
<|autheur|>:
|
21 |
+
- human
|
22 |
+
<|khey|>:
|
23 |
+
- gpt
|
24 |
+
<|sujet|>:
|
25 |
+
- system
|
26 |
+
dataset_prepared_path:
|
27 |
+
val_set_size: 0.05
|
28 |
+
output_dir: ./outputs/
|
29 |
+
dataset_prepared_path: last_run_prepared
|
30 |
+
|
31 |
+
sequence_len: 8192
|
32 |
+
sample_packing: true
|
33 |
+
pad_to_sequence_len: true
|
34 |
+
|
35 |
+
wandb_project: JVCGPT Medium 8b v2
|
36 |
+
wandb_entity:
|
37 |
+
wandb_watch:
|
38 |
+
wandb_name:
|
39 |
+
wandb_log_model:
|
40 |
+
|
41 |
+
gradient_accumulation_steps: 1
|
42 |
+
micro_batch_size: 1
|
43 |
+
num_epochs: 2
|
44 |
+
optimizer: paged_adamw_8bit
|
45 |
+
lr_scheduler: cosine
|
46 |
+
learning_rate: 1e-5
|
47 |
+
|
48 |
+
train_on_inputs: true
|
49 |
+
group_by_length: false
|
50 |
+
bf16: auto
|
51 |
+
fp16:
|
52 |
+
tf32: false
|
53 |
+
|
54 |
+
gradient_checkpointing: unsloth
|
55 |
+
gradient_checkpointing_kwargs:
|
56 |
+
use_reentrant: false
|
57 |
+
early_stopping_patience:
|
58 |
+
resume_from_checkpoint:
|
59 |
+
local_rank:
|
60 |
+
logging_steps: 1
|
61 |
+
xformers_attention:
|
62 |
+
flash_attention: true
|
63 |
+
s2_attention:
|
64 |
+
|
65 |
+
warmup_steps: 100
|
66 |
+
eval_table_size:
|
67 |
+
saves_per_epoch: 20
|
68 |
+
debug:
|
69 |
+
deepspeed:
|
70 |
+
weight_decay: 0.0
|
71 |
+
fsdp:
|
72 |
+
fsdp_config:
|
73 |
+
special_tokens:
|
74 |
+
pad_token: <|end_of_text|>
|
75 |
+
save_safetensors: true
|
76 |
+
save_total_limit: 5
|
77 |
+
resume_from_checkpoint: ./outputs/checkpoint-9999
|
78 |
+
# If resume_from_checkpoint isn't set and you simply want it to start where it left off.
|
79 |
+
# Be careful with this being turned on between different models.
|
80 |
+
auto_resume_from_checkpoints: true
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00000-of-00044.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:ddfab693fc86f01798957056e6fa5021e6fb3f078cf272ce47881420dba44c48
|
3 |
+
size 509510664
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00001-of-00044.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:efad90b6e1fa34cf8fb2d6df330a76932e3faaf12d544690ce1cdcca9be296ea
|
3 |
+
size 510345816
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00002-of-00044.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:5a5ffbc3d6dbaad5d094085b00c117994d1ca85acb491181d8ee61af76e22773
|
3 |
+
size 513083152
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00003-of-00044.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:aa01ed317b5aca4a0c237785f46312d19f74c02be8ed8604f999204b235f8916
|
3 |
+
size 492867696
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00004-of-00044.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:1db41ed88f689ed4178e943548f3f795d8091d1889e3d618b4abf1db39c964a0
|
3 |
+
size 502555736
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00005-of-00044.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:db62916b41a8d4b1412c4af753421f805ec0908fec42b5114e4fa220c6928bf6
|
3 |
+
size 516054648
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00006-of-00044.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:6c7e3b519dd59f408ea3e4879540b6e0db597bf322b4c147125c35005bd71929
|
3 |
+
size 504759640
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00007-of-00044.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:c8a7689eac1b75541a4a8ccac43d5505cda08d2213420cea85f9d056683b730a
|
3 |
+
size 497994960
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00008-of-00044.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:b7a87d5f57b224804b83acad441c834cdd9991784bb960215484fc7bbaa2f7dc
|
3 |
+
size 493499776
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00009-of-00044.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:c58edf651b900a1e32bea603d7835759aeead15e10cec7f8112301ef05dd3de1
|
3 |
+
size 492206160
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00010-of-00044.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:e77652ced005c44787b058887dd1771c5c2db228fe780561afe6f214ab1833fd
|
3 |
+
size 489879488
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00011-of-00044.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:af654bbe262b37e567a3304dd5a109a0c6483928f03aac9bf2a60fdb603637ea
|
3 |
+
size 515715544
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00012-of-00044.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:33ab10567511f8bd08ef9508fdf43d39cac1b5347f56a2b8015f0027225a3dfa
|
3 |
+
size 503197448
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00013-of-00044.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:c9d619a9e14ad590a83165575405404252902bb036061c428357a37d149d0cd7
|
3 |
+
size 495734056
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00014-of-00044.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:4e078bd874561204204d290e5dcf23cd4347ca3e51786cc7479cd1b9b64641c4
|
3 |
+
size 507576336
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00015-of-00044.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:579c3b55ad9406c35a71d98d92ff2dd0f9680f74be09762ab64d893a15879e5f
|
3 |
+
size 492115360
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00016-of-00044.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:b2f36d614a926b750af853324800886fd7f2856178dc0452f340fea155c75cc0
|
3 |
+
size 507268696
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00017-of-00044.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:3c5453b4dcd47679da5451212b2b3129ef50a36d25655af01f15c0b03b1d5171
|
3 |
+
size 499077744
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00018-of-00044.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:b7c97ed9786aa331c00ff4c1e7140343d4c3a610aaa34355b051a935f4acde1b
|
3 |
+
size 465142808
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00019-of-00044.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:805225a196da76f21c7cf8e7ac4e44a947706150031fbb09ad5363687a115233
|
3 |
+
size 465576552
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00020-of-00044.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:08d9b8ae8c537cb36757e46a32ab35acc934294306663549daa8c30cd9619f07
|
3 |
+
size 481875960
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00021-of-00044.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:0dd9573717340933049cb247238d002e3788f8e02d327675ea6d4268e521afc9
|
3 |
+
size 487433776
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00022-of-00044.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:6b0f8f6e6de402888fc298ed33c10dc12fb6184f6380d7a722a72577927ea4b6
|
3 |
+
size 469827672
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00023-of-00044.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:a63b558562ced2a6c1bf2992695c6f7d42c04cea1436ee46a8a30d6518908c24
|
3 |
+
size 491439744
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00024-of-00044.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:9fae272e02c85832ceeb6b3c4bb5764daec1f82d32a55290207bb47bd1fb84b5
|
3 |
+
size 486457504
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00025-of-00044.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:70a18c904fe752b676f7aa562943c33e49b6835473da7ec88dece23a95ffdfc7
|
3 |
+
size 506509512
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00026-of-00044.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:7be9c56ed965bab455770e664fe58c91b2119ff9b8b270b3d2b1c0d7c1183f19
|
3 |
+
size 493793960
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00027-of-00044.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:3bee33d5ad7e6651d9d5c9e3312ccc3517772ccc10ae446fe06e6992c1a5a551
|
3 |
+
size 492451264
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00028-of-00044.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:b6a7cd0288cc7ef1a8a673f3bc619f9a8503babb0ff9e0a160ce60b9c5818678
|
3 |
+
size 484112264
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00029-of-00044.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:102fc0ca1df843ef206e1ecf4b1a7dbda9585fcb849ae1ae471bd08ed104314e
|
3 |
+
size 483738896
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00030-of-00044.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:6721e3b7209976849b20b004a624a63ff3295eaea9a3572716f8bb5d2ba4792e
|
3 |
+
size 495332264
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00031-of-00044.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:a585e992b2bc601a21c09344a2206908b1a70f50fadcd59ceea1cdbbc54c613e
|
3 |
+
size 490535760
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00032-of-00044.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:cca1a1fe9b2d6024e9ce1e715dbf54101a850658510716e3f5090072f515544c
|
3 |
+
size 488630664
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00033-of-00044.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:6d6a693774eb3e63333469af775b613bc8428a513b6f91280b48678d5674eb49
|
3 |
+
size 482262112
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00034-of-00044.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:14cd9e6734f34f4bc05dc714f64436668f36b7145c4a7416fabc91c7ab31309f
|
3 |
+
size 485965880
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00035-of-00044.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:3fc0df1fec41b34463f7530dad39d76e1a36af45bee49a387ff92b155ceb6915
|
3 |
+
size 493787256
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00036-of-00044.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:07f5cc37a2d2586b9a3cb99ad6e72cf75b663816d99311e24d7d9b079d051111
|
3 |
+
size 496353552
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00037-of-00044.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:4240c83ff2c734e1476369fc259d2bbe218203f0b5b4f2c24d4e62c14d791f7c
|
3 |
+
size 499160384
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00038-of-00044.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:187c09cecb1c10bbfec3f3360b253b8bc49ad4410159a7f9c05417d40afca94c
|
3 |
+
size 508614656
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00039-of-00044.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:08a29063ca2b9c30db7a42b8d58ce7bb8aee76183a6323c4eb6b3f62c23b3f10
|
3 |
+
size 511701496
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00040-of-00044.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:373eaaf5dca5c00d6d28f305c836a9f3d9cbceeaed7ff3b057f8e2a05535078f
|
3 |
+
size 488775304
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00041-of-00044.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:b97dc36c49cfefee0308738020ff268e1a96ccd6cf21050268c52b136fedc0f9
|
3 |
+
size 500172336
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00042-of-00044.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:bb77f331d4664d96d16a7b417e35c81e03f0c3b7044e9f5c7bf442326fefcab4
|
3 |
+
size 496506336
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/data-00043-of-00044.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:290d4d35cb31b800ec008f3348dafc15f1fe5c78d2b66f824749fd238edcaae7
|
3 |
+
size 491838512
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/dataset_info.json
ADDED
@@ -0,0 +1,40 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"citation": "",
|
3 |
+
"description": "",
|
4 |
+
"features": {
|
5 |
+
"input_ids": {
|
6 |
+
"feature": {
|
7 |
+
"dtype": "int32",
|
8 |
+
"_type": "Value"
|
9 |
+
},
|
10 |
+
"_type": "Sequence"
|
11 |
+
},
|
12 |
+
"attention_mask": {
|
13 |
+
"feature": {
|
14 |
+
"dtype": "int8",
|
15 |
+
"_type": "Value"
|
16 |
+
},
|
17 |
+
"_type": "Sequence"
|
18 |
+
},
|
19 |
+
"labels": {
|
20 |
+
"feature": {
|
21 |
+
"dtype": "int64",
|
22 |
+
"_type": "Value"
|
23 |
+
},
|
24 |
+
"_type": "Sequence"
|
25 |
+
},
|
26 |
+
"position_ids": {
|
27 |
+
"feature": {
|
28 |
+
"dtype": "int64",
|
29 |
+
"_type": "Value"
|
30 |
+
},
|
31 |
+
"_type": "Sequence"
|
32 |
+
},
|
33 |
+
"length": {
|
34 |
+
"dtype": "int64",
|
35 |
+
"_type": "Value"
|
36 |
+
}
|
37 |
+
},
|
38 |
+
"homepage": "",
|
39 |
+
"license": ""
|
40 |
+
}
|
last_run_prepared/369c2adbdf251cc1440f6c85e7727ed6/state.json
ADDED
@@ -0,0 +1,142 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_data_files": [
|
3 |
+
{
|
4 |
+
"filename": "data-00000-of-00044.arrow"
|
5 |
+
},
|
6 |
+
{
|
7 |
+
"filename": "data-00001-of-00044.arrow"
|
8 |
+
},
|
9 |
+
{
|
10 |
+
"filename": "data-00002-of-00044.arrow"
|
11 |
+
},
|
12 |
+
{
|
13 |
+
"filename": "data-00003-of-00044.arrow"
|
14 |
+
},
|
15 |
+
{
|
16 |
+
"filename": "data-00004-of-00044.arrow"
|
17 |
+
},
|
18 |
+
{
|
19 |
+
"filename": "data-00005-of-00044.arrow"
|
20 |
+
},
|
21 |
+
{
|
22 |
+
"filename": "data-00006-of-00044.arrow"
|
23 |
+
},
|
24 |
+
{
|
25 |
+
"filename": "data-00007-of-00044.arrow"
|
26 |
+
},
|
27 |
+
{
|
28 |
+
"filename": "data-00008-of-00044.arrow"
|
29 |
+
},
|
30 |
+
{
|
31 |
+
"filename": "data-00009-of-00044.arrow"
|
32 |
+
},
|
33 |
+
{
|
34 |
+
"filename": "data-00010-of-00044.arrow"
|
35 |
+
},
|
36 |
+
{
|
37 |
+
"filename": "data-00011-of-00044.arrow"
|
38 |
+
},
|
39 |
+
{
|
40 |
+
"filename": "data-00012-of-00044.arrow"
|
41 |
+
},
|
42 |
+
{
|
43 |
+
"filename": "data-00013-of-00044.arrow"
|
44 |
+
},
|
45 |
+
{
|
46 |
+
"filename": "data-00014-of-00044.arrow"
|
47 |
+
},
|
48 |
+
{
|
49 |
+
"filename": "data-00015-of-00044.arrow"
|
50 |
+
},
|
51 |
+
{
|
52 |
+
"filename": "data-00016-of-00044.arrow"
|
53 |
+
},
|
54 |
+
{
|
55 |
+
"filename": "data-00017-of-00044.arrow"
|
56 |
+
},
|
57 |
+
{
|
58 |
+
"filename": "data-00018-of-00044.arrow"
|
59 |
+
},
|
60 |
+
{
|
61 |
+
"filename": "data-00019-of-00044.arrow"
|
62 |
+
},
|
63 |
+
{
|
64 |
+
"filename": "data-00020-of-00044.arrow"
|
65 |
+
},
|
66 |
+
{
|
67 |
+
"filename": "data-00021-of-00044.arrow"
|
68 |
+
},
|
69 |
+
{
|
70 |
+
"filename": "data-00022-of-00044.arrow"
|
71 |
+
},
|
72 |
+
{
|
73 |
+
"filename": "data-00023-of-00044.arrow"
|
74 |
+
},
|
75 |
+
{
|
76 |
+
"filename": "data-00024-of-00044.arrow"
|
77 |
+
},
|
78 |
+
{
|
79 |
+
"filename": "data-00025-of-00044.arrow"
|
80 |
+
},
|
81 |
+
{
|
82 |
+
"filename": "data-00026-of-00044.arrow"
|
83 |
+
},
|
84 |
+
{
|
85 |
+
"filename": "data-00027-of-00044.arrow"
|
86 |
+
},
|
87 |
+
{
|
88 |
+
"filename": "data-00028-of-00044.arrow"
|
89 |
+
},
|
90 |
+
{
|
91 |
+
"filename": "data-00029-of-00044.arrow"
|
92 |
+
},
|
93 |
+
{
|
94 |
+
"filename": "data-00030-of-00044.arrow"
|
95 |
+
},
|
96 |
+
{
|
97 |
+
"filename": "data-00031-of-00044.arrow"
|
98 |
+
},
|
99 |
+
{
|
100 |
+
"filename": "data-00032-of-00044.arrow"
|
101 |
+
},
|
102 |
+
{
|
103 |
+
"filename": "data-00033-of-00044.arrow"
|
104 |
+
},
|
105 |
+
{
|
106 |
+
"filename": "data-00034-of-00044.arrow"
|
107 |
+
},
|
108 |
+
{
|
109 |
+
"filename": "data-00035-of-00044.arrow"
|
110 |
+
},
|
111 |
+
{
|
112 |
+
"filename": "data-00036-of-00044.arrow"
|
113 |
+
},
|
114 |
+
{
|
115 |
+
"filename": "data-00037-of-00044.arrow"
|
116 |
+
},
|
117 |
+
{
|
118 |
+
"filename": "data-00038-of-00044.arrow"
|
119 |
+
},
|
120 |
+
{
|
121 |
+
"filename": "data-00039-of-00044.arrow"
|
122 |
+
},
|
123 |
+
{
|
124 |
+
"filename": "data-00040-of-00044.arrow"
|
125 |
+
},
|
126 |
+
{
|
127 |
+
"filename": "data-00041-of-00044.arrow"
|
128 |
+
},
|
129 |
+
{
|
130 |
+
"filename": "data-00042-of-00044.arrow"
|
131 |
+
},
|
132 |
+
{
|
133 |
+
"filename": "data-00043-of-00044.arrow"
|
134 |
+
}
|
135 |
+
],
|
136 |
+
"_fingerprint": "62a4782ee0a6bc85",
|
137 |
+
"_format_columns": null,
|
138 |
+
"_format_kwargs": {},
|
139 |
+
"_format_type": null,
|
140 |
+
"_output_all_columns": false,
|
141 |
+
"_split": null
|
142 |
+
}
|