upload model outputs
Browse files- checkpoints/best.pt +3 -0
- checkpoints/last.pt +3 -0
- config_merged.yaml +61 -0
- loss_plots/lmask.png +0 -0
- loss_plots/lqa.png +0 -0
- loss_plots/lrank.png +0 -0
- loss_plots/total.png +0 -0
- metrics.json +655 -0
- run_summary.json +7 -0
checkpoints/best.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1603dc5bf965144116d0517d69fa42b4bcce72d95e9392fe5a4e2bad6496e473
|
| 3 |
+
size 202807263
|
checkpoints/last.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ac0d7a55b8b78ff5f99ca7a3e75aec8781d71385372af4f37c08916064d3ced2
|
| 3 |
+
size 202814239
|
config_merged.yaml
ADDED
|
@@ -0,0 +1,61 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
model:
|
| 2 |
+
vocab_size: 32768
|
| 3 |
+
hidden_size: 512
|
| 4 |
+
policy_hidden_size: 512
|
| 5 |
+
num_queries: 32
|
| 6 |
+
heads: 8
|
| 7 |
+
depth: 3
|
| 8 |
+
dropout: 0.1
|
| 9 |
+
perceiver_num_latents: 64
|
| 10 |
+
perceiver_depth: 4
|
| 11 |
+
perceiver_dim_head: 64
|
| 12 |
+
perceiver_heads: 8
|
| 13 |
+
perceiver_ff_mult: 4
|
| 14 |
+
perceiver_qk_norm: false
|
| 15 |
+
qt_injection_mode: prefix
|
| 16 |
+
qwen_name: Qwen/Qwen3-4B-Instruct-2507
|
| 17 |
+
qwen_dtype: bfloat16
|
| 18 |
+
qwen_max_len: 2048
|
| 19 |
+
freeze_qwen: true
|
| 20 |
+
qwen_gradient_checkpointing: false
|
| 21 |
+
vision_encoder_name: google/siglip-so400m-patch14-384
|
| 22 |
+
vision_dtype: bfloat16
|
| 23 |
+
freeze_vision: true
|
| 24 |
+
data:
|
| 25 |
+
train_jsonl: /home/xwh/processed/policy_qformer/qasper_policy_train_dev.jsonl
|
| 26 |
+
dev_jsonl: /home/xwh/processed/policy_qformer/qasper_policy_train_dev.jsonl
|
| 27 |
+
batch_size: 2
|
| 28 |
+
max_ct_len: 384
|
| 29 |
+
max_qt_len: 128
|
| 30 |
+
max_a_len: 96
|
| 31 |
+
num_workers: 0
|
| 32 |
+
tokenizer_name: Qwen/Qwen3-4B-Instruct-2507
|
| 33 |
+
truncation_strategy: head
|
| 34 |
+
image_size: 384
|
| 35 |
+
train:
|
| 36 |
+
stage_a_epochs: 25
|
| 37 |
+
stage_b_epochs: 25
|
| 38 |
+
early_stop_patience: 1000
|
| 39 |
+
eval_every_epoch: true
|
| 40 |
+
save_every_epoch: true
|
| 41 |
+
loss:
|
| 42 |
+
alpha_mask: 0.3
|
| 43 |
+
beta_rank: 0.1
|
| 44 |
+
alpha_mask_warmup_epochs: 0
|
| 45 |
+
beta_rank_warmup_epochs: 0
|
| 46 |
+
optim:
|
| 47 |
+
lr: 0.0001
|
| 48 |
+
weight_decay: 0.01
|
| 49 |
+
grad_accum_steps: 1
|
| 50 |
+
scheduler: constant
|
| 51 |
+
warmup_steps: 0
|
| 52 |
+
runtime:
|
| 53 |
+
device: cuda
|
| 54 |
+
bf16: true
|
| 55 |
+
seed: 42
|
| 56 |
+
io:
|
| 57 |
+
output_dir: /home/xwh/qformer/outputs/perceiver_qasper_full
|
| 58 |
+
exp_name: policy_qformer_default
|
| 59 |
+
inference:
|
| 60 |
+
use_vllm: true
|
| 61 |
+
vllm_tensor_parallel_size: 1
|
loss_plots/lmask.png
ADDED
|
loss_plots/lqa.png
ADDED
|
loss_plots/lrank.png
ADDED
|
loss_plots/total.png
ADDED
|
metrics.json
ADDED
|
@@ -0,0 +1,655 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_score": 0.6681005201442157,
|
| 3 |
+
"history": [
|
| 4 |
+
{
|
| 5 |
+
"event": "eval_epoch",
|
| 6 |
+
"stage": "A",
|
| 7 |
+
"epoch": 1,
|
| 8 |
+
"global_step": 1050,
|
| 9 |
+
"qa_token_acc": 0.0026088659046127876,
|
| 10 |
+
"qa_em": 0.0,
|
| 11 |
+
"qa_f1": 0.008460056155250058,
|
| 12 |
+
"rank_r1": 0.9179775280898876,
|
| 13 |
+
"rank_r5": 1.0,
|
| 14 |
+
"rank_mrr": 0.9589887640449438,
|
| 15 |
+
"dev_score": 0.4837244101000969
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"event": "eval_epoch",
|
| 19 |
+
"stage": "A",
|
| 20 |
+
"epoch": 2,
|
| 21 |
+
"global_step": 2100,
|
| 22 |
+
"qa_token_acc": 0.0002740576395289882,
|
| 23 |
+
"qa_em": 0.0,
|
| 24 |
+
"qa_f1": 0.0021882260907039866,
|
| 25 |
+
"rank_r1": 0.9235955056179775,
|
| 26 |
+
"rank_r5": 1.0,
|
| 27 |
+
"rank_mrr": 0.9617977528089887,
|
| 28 |
+
"dev_score": 0.48199298944984637
|
| 29 |
+
},
|
| 30 |
+
{
|
| 31 |
+
"event": "eval_epoch",
|
| 32 |
+
"stage": "A",
|
| 33 |
+
"epoch": 3,
|
| 34 |
+
"global_step": 3150,
|
| 35 |
+
"qa_token_acc": 0.002442889937898036,
|
| 36 |
+
"qa_em": 0.0,
|
| 37 |
+
"qa_f1": 0.010146229290939182,
|
| 38 |
+
"rank_r1": 0.9258426966292135,
|
| 39 |
+
"rank_r5": 1.0,
|
| 40 |
+
"rank_mrr": 0.9629213483146067,
|
| 41 |
+
"dev_score": 0.486533788802773
|
| 42 |
+
},
|
| 43 |
+
{
|
| 44 |
+
"event": "eval_epoch",
|
| 45 |
+
"stage": "A",
|
| 46 |
+
"epoch": 4,
|
| 47 |
+
"global_step": 4200,
|
| 48 |
+
"qa_token_acc": 0.002491740051239843,
|
| 49 |
+
"qa_em": 0.0,
|
| 50 |
+
"qa_f1": 0.00798511925920835,
|
| 51 |
+
"rank_r1": 0.9089887640449438,
|
| 52 |
+
"rank_r5": 1.0,
|
| 53 |
+
"rank_mrr": 0.9544943820224719,
|
| 54 |
+
"dev_score": 0.4812397506408401
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"event": "eval_epoch",
|
| 58 |
+
"stage": "A",
|
| 59 |
+
"epoch": 5,
|
| 60 |
+
"global_step": 5250,
|
| 61 |
+
"qa_token_acc": 0.0021886772463160474,
|
| 62 |
+
"qa_em": 0.0,
|
| 63 |
+
"qa_f1": 0.018776815370558517,
|
| 64 |
+
"rank_r1": 0.946067415730337,
|
| 65 |
+
"rank_r5": 1.0,
|
| 66 |
+
"rank_mrr": 0.9730337078651685,
|
| 67 |
+
"dev_score": 0.4959052616178635
|
| 68 |
+
},
|
| 69 |
+
{
|
| 70 |
+
"event": "eval_epoch",
|
| 71 |
+
"stage": "A",
|
| 72 |
+
"epoch": 6,
|
| 73 |
+
"global_step": 6300,
|
| 74 |
+
"qa_token_acc": 0.00150491853246987,
|
| 75 |
+
"qa_em": 0.0,
|
| 76 |
+
"qa_f1": 0.010527824174669367,
|
| 77 |
+
"rank_r1": 0.9550561797752809,
|
| 78 |
+
"rank_r5": 1.0,
|
| 79 |
+
"rank_mrr": 0.9775280898876404,
|
| 80 |
+
"dev_score": 0.4940279570311549
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"event": "eval_epoch",
|
| 84 |
+
"stage": "A",
|
| 85 |
+
"epoch": 7,
|
| 86 |
+
"global_step": 7350,
|
| 87 |
+
"qa_token_acc": 0.0007350574409100871,
|
| 88 |
+
"qa_em": 0.0,
|
| 89 |
+
"qa_f1": 0.005656881949104641,
|
| 90 |
+
"rank_r1": 0.9382022471910112,
|
| 91 |
+
"rank_r5": 1.0,
|
| 92 |
+
"rank_mrr": 0.9691011235955056,
|
| 93 |
+
"dev_score": 0.4873790027723051
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"event": "eval_epoch",
|
| 97 |
+
"stage": "A",
|
| 98 |
+
"epoch": 8,
|
| 99 |
+
"global_step": 8400,
|
| 100 |
+
"qa_token_acc": 0.0019801811067466503,
|
| 101 |
+
"qa_em": 0.0,
|
| 102 |
+
"qa_f1": 0.016576542164749433,
|
| 103 |
+
"rank_r1": 0.9426966292134832,
|
| 104 |
+
"rank_r5": 1.0,
|
| 105 |
+
"rank_mrr": 0.9713483146067415,
|
| 106 |
+
"dev_score": 0.4939624283857455
|
| 107 |
+
},
|
| 108 |
+
{
|
| 109 |
+
"event": "eval_epoch",
|
| 110 |
+
"stage": "A",
|
| 111 |
+
"epoch": 9,
|
| 112 |
+
"global_step": 9450,
|
| 113 |
+
"qa_token_acc": 0.003337956813784035,
|
| 114 |
+
"qa_em": 0.0,
|
| 115 |
+
"qa_f1": 0.027663325242331938,
|
| 116 |
+
"rank_r1": 0.950561797752809,
|
| 117 |
+
"rank_r5": 1.0,
|
| 118 |
+
"rank_mrr": 0.9752808988764045,
|
| 119 |
+
"dev_score": 0.5014721120593681
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"event": "eval_epoch",
|
| 123 |
+
"stage": "A",
|
| 124 |
+
"epoch": 10,
|
| 125 |
+
"global_step": 10500,
|
| 126 |
+
"qa_token_acc": 0.002120477749868174,
|
| 127 |
+
"qa_em": 0.0,
|
| 128 |
+
"qa_f1": 0.015086279129733322,
|
| 129 |
+
"rank_r1": 0.9348314606741573,
|
| 130 |
+
"rank_r5": 1.0,
|
| 131 |
+
"rank_mrr": 0.9674157303370786,
|
| 132 |
+
"dev_score": 0.491251004733406
|
| 133 |
+
},
|
| 134 |
+
{
|
| 135 |
+
"event": "eval_epoch",
|
| 136 |
+
"stage": "A",
|
| 137 |
+
"epoch": 11,
|
| 138 |
+
"global_step": 11550,
|
| 139 |
+
"qa_token_acc": 0.004011323542787213,
|
| 140 |
+
"qa_em": 0.0,
|
| 141 |
+
"qa_f1": 0.021823077741655643,
|
| 142 |
+
"rank_r1": 0.9382022471910112,
|
| 143 |
+
"rank_r5": 1.0,
|
| 144 |
+
"rank_mrr": 0.9691011235955056,
|
| 145 |
+
"dev_score": 0.4954621006685806
|
| 146 |
+
},
|
| 147 |
+
{
|
| 148 |
+
"event": "eval_epoch",
|
| 149 |
+
"stage": "A",
|
| 150 |
+
"epoch": 12,
|
| 151 |
+
"global_step": 12600,
|
| 152 |
+
"qa_token_acc": 0.0020072296711053372,
|
| 153 |
+
"qa_em": 0.0,
|
| 154 |
+
"qa_f1": 0.019076629476642668,
|
| 155 |
+
"rank_r1": 0.946067415730337,
|
| 156 |
+
"rank_r5": 1.0,
|
| 157 |
+
"rank_mrr": 0.9730337078651685,
|
| 158 |
+
"dev_score": 0.4960551686709056
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"event": "eval_epoch",
|
| 162 |
+
"stage": "A",
|
| 163 |
+
"epoch": 13,
|
| 164 |
+
"global_step": 13650,
|
| 165 |
+
"qa_token_acc": 0.004285261014129599,
|
| 166 |
+
"qa_em": 0.0,
|
| 167 |
+
"qa_f1": 0.02629726631797427,
|
| 168 |
+
"rank_r1": 0.9269662921348315,
|
| 169 |
+
"rank_r5": 1.0,
|
| 170 |
+
"rank_mrr": 0.9634831460674157,
|
| 171 |
+
"dev_score": 0.49489020619269497
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"event": "eval_epoch",
|
| 175 |
+
"stage": "A",
|
| 176 |
+
"epoch": 14,
|
| 177 |
+
"global_step": 14700,
|
| 178 |
+
"qa_token_acc": 0.0021624533130300636,
|
| 179 |
+
"qa_em": 0.0,
|
| 180 |
+
"qa_f1": 0.013479208688918377,
|
| 181 |
+
"rank_r1": 0.9325842696629213,
|
| 182 |
+
"rank_r5": 1.0,
|
| 183 |
+
"rank_mrr": 0.9662921348314607,
|
| 184 |
+
"dev_score": 0.48988567176018954
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"event": "eval_epoch",
|
| 188 |
+
"stage": "A",
|
| 189 |
+
"epoch": 15,
|
| 190 |
+
"global_step": 15750,
|
| 191 |
+
"qa_token_acc": 0.0017375697114891482,
|
| 192 |
+
"qa_em": 0.0,
|
| 193 |
+
"qa_f1": 0.010337309586139602,
|
| 194 |
+
"rank_r1": 0.9595505617977528,
|
| 195 |
+
"rank_r5": 1.0,
|
| 196 |
+
"rank_mrr": 0.9797752808988764,
|
| 197 |
+
"dev_score": 0.49505629524250805
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"event": "eval_epoch",
|
| 201 |
+
"stage": "A",
|
| 202 |
+
"epoch": 16,
|
| 203 |
+
"global_step": 16800,
|
| 204 |
+
"qa_token_acc": 0.0031858088179451674,
|
| 205 |
+
"qa_em": 0.0,
|
| 206 |
+
"qa_f1": 0.010815200878153929,
|
| 207 |
+
"rank_r1": 0.9561797752808989,
|
| 208 |
+
"rank_r5": 1.0,
|
| 209 |
+
"rank_mrr": 0.9780898876404495,
|
| 210 |
+
"dev_score": 0.4944525442593017
|
| 211 |
+
},
|
| 212 |
+
{
|
| 213 |
+
"event": "eval_epoch",
|
| 214 |
+
"stage": "A",
|
| 215 |
+
"epoch": 17,
|
| 216 |
+
"global_step": 17850,
|
| 217 |
+
"qa_token_acc": 0.002876572579325557,
|
| 218 |
+
"qa_em": 0.0,
|
| 219 |
+
"qa_f1": 0.018080411167543484,
|
| 220 |
+
"rank_r1": 0.9539325842696629,
|
| 221 |
+
"rank_r5": 1.0,
|
| 222 |
+
"rank_mrr": 0.9769662921348314,
|
| 223 |
+
"dev_score": 0.49752335165118744
|
| 224 |
+
},
|
| 225 |
+
{
|
| 226 |
+
"event": "eval_epoch",
|
| 227 |
+
"stage": "A",
|
| 228 |
+
"epoch": 18,
|
| 229 |
+
"global_step": 18900,
|
| 230 |
+
"qa_token_acc": 0.005014380686699785,
|
| 231 |
+
"qa_em": 0.0,
|
| 232 |
+
"qa_f1": 0.01655073734085142,
|
| 233 |
+
"rank_r1": 0.949438202247191,
|
| 234 |
+
"rank_r5": 1.0,
|
| 235 |
+
"rank_mrr": 0.9747191011235955,
|
| 236 |
+
"dev_score": 0.49563491923222347
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"event": "eval_epoch",
|
| 240 |
+
"stage": "A",
|
| 241 |
+
"epoch": 19,
|
| 242 |
+
"global_step": 19950,
|
| 243 |
+
"qa_token_acc": 0.0024065798941515816,
|
| 244 |
+
"qa_em": 0.0,
|
| 245 |
+
"qa_f1": 0.01064849724661803,
|
| 246 |
+
"rank_r1": 0.9561797752808989,
|
| 247 |
+
"rank_r5": 1.0,
|
| 248 |
+
"rank_mrr": 0.9780898876404495,
|
| 249 |
+
"dev_score": 0.4943691924435337
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"event": "eval_epoch",
|
| 253 |
+
"stage": "A",
|
| 254 |
+
"epoch": 20,
|
| 255 |
+
"global_step": 21000,
|
| 256 |
+
"qa_token_acc": 0.001667128692262136,
|
| 257 |
+
"qa_em": 0.0,
|
| 258 |
+
"qa_f1": 0.011459056958696605,
|
| 259 |
+
"rank_r1": 0.9550561797752809,
|
| 260 |
+
"rank_r5": 1.0,
|
| 261 |
+
"rank_mrr": 0.9775280898876404,
|
| 262 |
+
"dev_score": 0.4944935734231685
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"event": "eval_epoch",
|
| 266 |
+
"stage": "A",
|
| 267 |
+
"epoch": 21,
|
| 268 |
+
"global_step": 22050,
|
| 269 |
+
"qa_token_acc": 0.0010081766771694268,
|
| 270 |
+
"qa_em": 0.0,
|
| 271 |
+
"qa_f1": 0.010952496004514445,
|
| 272 |
+
"rank_r1": 0.9640449438202248,
|
| 273 |
+
"rank_r5": 1.0,
|
| 274 |
+
"rank_mrr": 0.9820224719101124,
|
| 275 |
+
"dev_score": 0.4964874839573134
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"event": "eval_epoch",
|
| 279 |
+
"stage": "A",
|
| 280 |
+
"epoch": 22,
|
| 281 |
+
"global_step": 23100,
|
| 282 |
+
"qa_token_acc": 0.0067095332047939745,
|
| 283 |
+
"qa_em": 0.0,
|
| 284 |
+
"qa_f1": 0.02894461489169904,
|
| 285 |
+
"rank_r1": 0.950561797752809,
|
| 286 |
+
"rank_r5": 1.0,
|
| 287 |
+
"rank_mrr": 0.9752808988764045,
|
| 288 |
+
"dev_score": 0.5021127568840518
|
| 289 |
+
},
|
| 290 |
+
{
|
| 291 |
+
"event": "eval_epoch",
|
| 292 |
+
"stage": "A",
|
| 293 |
+
"epoch": 23,
|
| 294 |
+
"global_step": 24150,
|
| 295 |
+
"qa_token_acc": 0.002842412927427731,
|
| 296 |
+
"qa_em": 0.0,
|
| 297 |
+
"qa_f1": 0.015541875253267664,
|
| 298 |
+
"rank_r1": 0.9651685393258427,
|
| 299 |
+
"rank_r5": 1.0,
|
| 300 |
+
"rank_mrr": 0.9825842696629213,
|
| 301 |
+
"dev_score": 0.4990630724580945
|
| 302 |
+
},
|
| 303 |
+
{
|
| 304 |
+
"event": "eval_epoch",
|
| 305 |
+
"stage": "A",
|
| 306 |
+
"epoch": 24,
|
| 307 |
+
"global_step": 25200,
|
| 308 |
+
"qa_token_acc": 0.0038040943512142413,
|
| 309 |
+
"qa_em": 0.0,
|
| 310 |
+
"qa_f1": 0.01523451581945387,
|
| 311 |
+
"rank_r1": 0.9224719101123595,
|
| 312 |
+
"rank_r5": 1.0,
|
| 313 |
+
"rank_mrr": 0.9612359550561798,
|
| 314 |
+
"dev_score": 0.48823523543781683
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"event": "eval_epoch",
|
| 318 |
+
"stage": "A",
|
| 319 |
+
"epoch": 25,
|
| 320 |
+
"global_step": 26250,
|
| 321 |
+
"qa_token_acc": 0.0020634263952415146,
|
| 322 |
+
"qa_em": 0.0,
|
| 323 |
+
"qa_f1": 0.006874889036257684,
|
| 324 |
+
"rank_r1": 0.9539325842696629,
|
| 325 |
+
"rank_r5": 1.0,
|
| 326 |
+
"rank_mrr": 0.9769662921348314,
|
| 327 |
+
"dev_score": 0.49192059058554455
|
| 328 |
+
},
|
| 329 |
+
{
|
| 330 |
+
"event": "eval_epoch",
|
| 331 |
+
"stage": "B",
|
| 332 |
+
"epoch": 1,
|
| 333 |
+
"global_step": 27300,
|
| 334 |
+
"qa_token_acc": 0.03595962160347243,
|
| 335 |
+
"qa_em": 0.0,
|
| 336 |
+
"qa_f1": 0.03595962160347243,
|
| 337 |
+
"rank_r1": 0.9595505617977528,
|
| 338 |
+
"rank_r5": 1.0,
|
| 339 |
+
"rank_mrr": 0.9797752808988764,
|
| 340 |
+
"dev_score": 0.5078674512511744
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"event": "eval_epoch",
|
| 344 |
+
"stage": "B",
|
| 345 |
+
"epoch": 2,
|
| 346 |
+
"global_step": 28350,
|
| 347 |
+
"qa_token_acc": 0.06349278112664943,
|
| 348 |
+
"qa_em": 0.0,
|
| 349 |
+
"qa_f1": 0.06402782542622525,
|
| 350 |
+
"rank_r1": 0.952808988764045,
|
| 351 |
+
"rank_r5": 1.0,
|
| 352 |
+
"rank_mrr": 0.9764044943820225,
|
| 353 |
+
"dev_score": 0.5202161599041238
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"event": "eval_epoch",
|
| 357 |
+
"stage": "B",
|
| 358 |
+
"epoch": 3,
|
| 359 |
+
"global_step": 29400,
|
| 360 |
+
"qa_token_acc": 0.0644060876516836,
|
| 361 |
+
"qa_em": 0.0,
|
| 362 |
+
"qa_f1": 0.06702022663240345,
|
| 363 |
+
"rank_r1": 0.9584269662921349,
|
| 364 |
+
"rank_r5": 1.0,
|
| 365 |
+
"rank_mrr": 0.9792134831460674,
|
| 366 |
+
"dev_score": 0.5231168548892354
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"event": "eval_epoch",
|
| 370 |
+
"stage": "B",
|
| 371 |
+
"epoch": 4,
|
| 372 |
+
"global_step": 30450,
|
| 373 |
+
"qa_token_acc": 0.06585100411920221,
|
| 374 |
+
"qa_em": 0.0,
|
| 375 |
+
"qa_f1": 0.06654645635210447,
|
| 376 |
+
"rank_r1": 0.9426966292134832,
|
| 377 |
+
"rank_r5": 1.0,
|
| 378 |
+
"rank_mrr": 0.9713483146067415,
|
| 379 |
+
"dev_score": 0.518947385479423
|
| 380 |
+
},
|
| 381 |
+
{
|
| 382 |
+
"event": "eval_epoch",
|
| 383 |
+
"stage": "B",
|
| 384 |
+
"epoch": 5,
|
| 385 |
+
"global_step": 31500,
|
| 386 |
+
"qa_token_acc": 0.06721366773564279,
|
| 387 |
+
"qa_em": 0.0,
|
| 388 |
+
"qa_f1": 0.0694767929892264,
|
| 389 |
+
"rank_r1": 0.949438202247191,
|
| 390 |
+
"rank_r5": 1.0,
|
| 391 |
+
"rank_mrr": 0.9747191011235955,
|
| 392 |
+
"dev_score": 0.522097947056411
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"event": "eval_epoch",
|
| 396 |
+
"stage": "B",
|
| 397 |
+
"epoch": 6,
|
| 398 |
+
"global_step": 32550,
|
| 399 |
+
"qa_token_acc": 0.06689300250838345,
|
| 400 |
+
"qa_em": 0.0,
|
| 401 |
+
"qa_f1": 0.06818316411067013,
|
| 402 |
+
"rank_r1": 0.9471910112359551,
|
| 403 |
+
"rank_r5": 1.0,
|
| 404 |
+
"rank_mrr": 0.9735955056179775,
|
| 405 |
+
"dev_score": 0.5208893348643238
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"event": "eval_epoch",
|
| 409 |
+
"stage": "B",
|
| 410 |
+
"epoch": 7,
|
| 411 |
+
"global_step": 33600,
|
| 412 |
+
"qa_token_acc": 0.06347600901661579,
|
| 413 |
+
"qa_em": 0.0022471910112359553,
|
| 414 |
+
"qa_f1": 0.06487733691557757,
|
| 415 |
+
"rank_r1": 0.9539325842696629,
|
| 416 |
+
"rank_r5": 1.0,
|
| 417 |
+
"rank_mrr": 0.9769662921348314,
|
| 418 |
+
"dev_score": 0.5209218145252045
|
| 419 |
+
},
|
| 420 |
+
{
|
| 421 |
+
"event": "eval_epoch",
|
| 422 |
+
"stage": "B",
|
| 423 |
+
"epoch": 8,
|
| 424 |
+
"global_step": 34650,
|
| 425 |
+
"qa_token_acc": 0.06677638813949753,
|
| 426 |
+
"qa_em": 0.0,
|
| 427 |
+
"qa_f1": 0.06790244536850619,
|
| 428 |
+
"rank_r1": 0.9539325842696629,
|
| 429 |
+
"rank_r5": 1.0,
|
| 430 |
+
"rank_mrr": 0.9769662921348314,
|
| 431 |
+
"dev_score": 0.5224343687516688
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"event": "eval_epoch",
|
| 435 |
+
"stage": "B",
|
| 436 |
+
"epoch": 9,
|
| 437 |
+
"global_step": 35700,
|
| 438 |
+
"qa_token_acc": 0.06686911011489442,
|
| 439 |
+
"qa_em": 0.0011235955056179776,
|
| 440 |
+
"qa_f1": 0.0683053846937451,
|
| 441 |
+
"rank_r1": 0.950561797752809,
|
| 442 |
+
"rank_r5": 1.0,
|
| 443 |
+
"rank_mrr": 0.9752808988764045,
|
| 444 |
+
"dev_score": 0.5217931417850747
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"event": "eval_epoch",
|
| 448 |
+
"stage": "B",
|
| 449 |
+
"epoch": 10,
|
| 450 |
+
"global_step": 36750,
|
| 451 |
+
"qa_token_acc": 0.061308408559934795,
|
| 452 |
+
"qa_em": 0.0,
|
| 453 |
+
"qa_f1": 0.06259742774850655,
|
| 454 |
+
"rank_r1": 0.9550561797752809,
|
| 455 |
+
"rank_r5": 1.0,
|
| 456 |
+
"rank_mrr": 0.9775280898876404,
|
| 457 |
+
"dev_score": 0.5200627588180735
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"event": "eval_epoch",
|
| 461 |
+
"stage": "B",
|
| 462 |
+
"epoch": 11,
|
| 463 |
+
"global_step": 37800,
|
| 464 |
+
"qa_token_acc": 0.06446662016820472,
|
| 465 |
+
"qa_em": 0.0,
|
| 466 |
+
"qa_f1": 0.06586377064292695,
|
| 467 |
+
"rank_r1": 0.949438202247191,
|
| 468 |
+
"rank_r5": 1.0,
|
| 469 |
+
"rank_mrr": 0.9747191011235955,
|
| 470 |
+
"dev_score": 0.5202914358832612
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"event": "eval_epoch",
|
| 474 |
+
"stage": "B",
|
| 475 |
+
"epoch": 12,
|
| 476 |
+
"global_step": 38850,
|
| 477 |
+
"qa_token_acc": 0.0664656804764573,
|
| 478 |
+
"qa_em": 0.0,
|
| 479 |
+
"qa_f1": 0.06790622950450839,
|
| 480 |
+
"rank_r1": 0.9550561797752809,
|
| 481 |
+
"rank_r5": 1.0,
|
| 482 |
+
"rank_mrr": 0.9775280898876404,
|
| 483 |
+
"dev_score": 0.5227171596960744
|
| 484 |
+
},
|
| 485 |
+
{
|
| 486 |
+
"event": "eval_epoch",
|
| 487 |
+
"stage": "B",
|
| 488 |
+
"epoch": 13,
|
| 489 |
+
"global_step": 39900,
|
| 490 |
+
"qa_token_acc": 0.06309952723736646,
|
| 491 |
+
"qa_em": 0.0022471910112359553,
|
| 492 |
+
"qa_f1": 0.06405629870588811,
|
| 493 |
+
"rank_r1": 0.949438202247191,
|
| 494 |
+
"rank_r5": 1.0,
|
| 495 |
+
"rank_mrr": 0.9747191011235955,
|
| 496 |
+
"dev_score": 0.5193876999147418
|
| 497 |
+
},
|
| 498 |
+
{
|
| 499 |
+
"event": "eval_epoch",
|
| 500 |
+
"stage": "B",
|
| 501 |
+
"epoch": 14,
|
| 502 |
+
"global_step": 40950,
|
| 503 |
+
"qa_token_acc": 0.06719656732612586,
|
| 504 |
+
"qa_em": 0.0,
|
| 505 |
+
"qa_f1": 0.06777480643511719,
|
| 506 |
+
"rank_r1": 0.950561797752809,
|
| 507 |
+
"rank_r5": 1.0,
|
| 508 |
+
"rank_mrr": 0.9752808988764045,
|
| 509 |
+
"dev_score": 0.5215278526557608
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"event": "eval_epoch",
|
| 513 |
+
"stage": "B",
|
| 514 |
+
"epoch": 15,
|
| 515 |
+
"global_step": 42000,
|
| 516 |
+
"qa_token_acc": 0.0670964259986907,
|
| 517 |
+
"qa_em": 0.0,
|
| 518 |
+
"qa_f1": 0.06829975286082439,
|
| 519 |
+
"rank_r1": 0.9573033707865168,
|
| 520 |
+
"rank_r5": 1.0,
|
| 521 |
+
"rank_mrr": 0.9786516853932584,
|
| 522 |
+
"dev_score": 0.5234757191270414
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"event": "eval_epoch",
|
| 526 |
+
"stage": "B",
|
| 527 |
+
"epoch": 16,
|
| 528 |
+
"global_step": 43050,
|
| 529 |
+
"qa_token_acc": 0.06484622750046914,
|
| 530 |
+
"qa_em": 0.007865168539325843,
|
| 531 |
+
"qa_f1": 0.06521137736749497,
|
| 532 |
+
"rank_r1": 0.9606741573033708,
|
| 533 |
+
"rank_r5": 1.0,
|
| 534 |
+
"rank_mrr": 0.9803370786516854,
|
| 535 |
+
"dev_score": 0.5227742280095902
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"event": "eval_epoch",
|
| 539 |
+
"stage": "B",
|
| 540 |
+
"epoch": 17,
|
| 541 |
+
"global_step": 44100,
|
| 542 |
+
"qa_token_acc": 0.07489982579562145,
|
| 543 |
+
"qa_em": 0.0044943820224719105,
|
| 544 |
+
"qa_f1": 0.07553913227954111,
|
| 545 |
+
"rank_r1": 0.9539325842696629,
|
| 546 |
+
"rank_r5": 1.0,
|
| 547 |
+
"rank_mrr": 0.9769662921348314,
|
| 548 |
+
"dev_score": 0.5262527122071863
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"event": "eval_epoch",
|
| 552 |
+
"stage": "B",
|
| 553 |
+
"epoch": 18,
|
| 554 |
+
"global_step": 45150,
|
| 555 |
+
"qa_token_acc": 0.0797927738995402,
|
| 556 |
+
"qa_em": 0.0022471910112359553,
|
| 557 |
+
"qa_f1": 0.08073487259905027,
|
| 558 |
+
"rank_r1": 0.9539325842696629,
|
| 559 |
+
"rank_r5": 1.0,
|
| 560 |
+
"rank_mrr": 0.9769662921348314,
|
| 561 |
+
"dev_score": 0.5288505823669408
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"event": "eval_epoch",
|
| 565 |
+
"stage": "B",
|
| 566 |
+
"epoch": 19,
|
| 567 |
+
"global_step": 46200,
|
| 568 |
+
"qa_token_acc": 0.08249868470286494,
|
| 569 |
+
"qa_em": 0.0022471910112359553,
|
| 570 |
+
"qa_f1": 0.08344526659295631,
|
| 571 |
+
"rank_r1": 0.9539325842696629,
|
| 572 |
+
"rank_r5": 1.0,
|
| 573 |
+
"rank_mrr": 0.9769662921348314,
|
| 574 |
+
"dev_score": 0.5302057793638939
|
| 575 |
+
},
|
| 576 |
+
{
|
| 577 |
+
"event": "eval_epoch",
|
| 578 |
+
"stage": "B",
|
| 579 |
+
"epoch": 20,
|
| 580 |
+
"global_step": 47250,
|
| 581 |
+
"qa_token_acc": 0.08896931529260833,
|
| 582 |
+
"qa_em": 0.0056179775280898875,
|
| 583 |
+
"qa_f1": 0.08979104532890575,
|
| 584 |
+
"rank_r1": 0.9606741573033708,
|
| 585 |
+
"rank_r5": 1.0,
|
| 586 |
+
"rank_mrr": 0.9803370786516854,
|
| 587 |
+
"dev_score": 0.5350640619902955
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"event": "eval_epoch",
|
| 591 |
+
"stage": "B",
|
| 592 |
+
"epoch": 21,
|
| 593 |
+
"global_step": 48300,
|
| 594 |
+
"qa_token_acc": 0.11453380242677842,
|
| 595 |
+
"qa_em": 0.010112359550561797,
|
| 596 |
+
"qa_f1": 0.11563916112067416,
|
| 597 |
+
"rank_r1": 0.949438202247191,
|
| 598 |
+
"rank_r5": 1.0,
|
| 599 |
+
"rank_mrr": 0.9747191011235955,
|
| 600 |
+
"dev_score": 0.5451791311221348
|
| 601 |
+
},
|
| 602 |
+
{
|
| 603 |
+
"event": "eval_epoch",
|
| 604 |
+
"stage": "B",
|
| 605 |
+
"epoch": 22,
|
| 606 |
+
"global_step": 49350,
|
| 607 |
+
"qa_token_acc": 0.13940242582300655,
|
| 608 |
+
"qa_em": 0.0044943820224719105,
|
| 609 |
+
"qa_f1": 0.14130278684123287,
|
| 610 |
+
"rank_r1": 0.9516853932584269,
|
| 611 |
+
"rank_r5": 1.0,
|
| 612 |
+
"rank_mrr": 0.9758426966292135,
|
| 613 |
+
"dev_score": 0.5585727417352232
|
| 614 |
+
},
|
| 615 |
+
{
|
| 616 |
+
"event": "eval_epoch",
|
| 617 |
+
"stage": "B",
|
| 618 |
+
"epoch": 23,
|
| 619 |
+
"global_step": 50400,
|
| 620 |
+
"qa_token_acc": 0.21105027623330114,
|
| 621 |
+
"qa_em": 0.012359550561797753,
|
| 622 |
+
"qa_f1": 0.2131529528282703,
|
| 623 |
+
"rank_r1": 0.949438202247191,
|
| 624 |
+
"rank_r5": 1.0,
|
| 625 |
+
"rank_mrr": 0.9747191011235955,
|
| 626 |
+
"dev_score": 0.5939360269759328
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"event": "eval_epoch",
|
| 630 |
+
"stage": "B",
|
| 631 |
+
"epoch": 24,
|
| 632 |
+
"global_step": 51450,
|
| 633 |
+
"qa_token_acc": 0.2707055699335783,
|
| 634 |
+
"qa_em": 0.017977528089887642,
|
| 635 |
+
"qa_f1": 0.27222918334792845,
|
| 636 |
+
"rank_r1": 0.952808988764045,
|
| 637 |
+
"rank_r5": 1.0,
|
| 638 |
+
"rank_mrr": 0.9764044943820225,
|
| 639 |
+
"dev_score": 0.6243168388649755
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"event": "eval_epoch",
|
| 643 |
+
"stage": "B",
|
| 644 |
+
"epoch": 25,
|
| 645 |
+
"global_step": 52500,
|
| 646 |
+
"qa_token_acc": 0.35802289772288703,
|
| 647 |
+
"qa_em": 0.019101123595505618,
|
| 648 |
+
"qa_f1": 0.35923474815360007,
|
| 649 |
+
"rank_r1": 0.9539325842696629,
|
| 650 |
+
"rank_r5": 1.0,
|
| 651 |
+
"rank_mrr": 0.9769662921348314,
|
| 652 |
+
"dev_score": 0.6681005201442157
|
| 653 |
+
}
|
| 654 |
+
]
|
| 655 |
+
}
|
run_summary.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"output_dir": "/home/xwh/qformer/outputs/perceiver_qasper_full",
|
| 3 |
+
"best_score": 0.6681005201442157,
|
| 4 |
+
"global_step": 52500,
|
| 5 |
+
"history_len": 50,
|
| 6 |
+
"tokenizer_pad_id": 151643
|
| 7 |
+
}
|