Feature Extraction
Transformers
TensorBoard
Safetensors
English
captionbert_v2
sentence-similarity
consensus-distillation
geometric-deep-learning
amoe
custom_code
Instructions to use AbstractPhil/captionbert-8192-v2 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use AbstractPhil/captionbert-8192-v2 with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("feature-extraction", model="AbstractPhil/captionbert-8192-v2", trust_remote_code=True)# pip install -U transformers accelerate # Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("AbstractPhil/captionbert-8192-v2", trust_remote_code=True, device_map="auto") - Notebooks
- Google Colab
- Kaggle
final: 2 anchors + dispatch + card
Browse files
amoe/collective/README.md
ADDED
|
@@ -0,0 +1,69 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: mit
|
| 3 |
+
tags: [amoe, adapter, mixture-of-experts, sentence-similarity, captionbert, aleph]
|
| 4 |
+
base_model: AbstractPhil/captionbert-8192-v2
|
| 5 |
+
library_name: amoe-lora
|
| 6 |
+
---
|
| 7 |
+
|
| 8 |
+
# captionbert-8192-v2 :: AMOE 2-anchor mixture
|
| 9 |
+
|
| 10 |
+
Two [amoe-lora](https://github.com/AbstractEyes/amoe-lora) aleph anchors on the
|
| 11 |
+
**frozen** trunk, plus a trained dispatch over them. The trunk never moves.
|
| 12 |
+
|
| 13 |
+
| anchor | trained on | relation |
|
| 14 |
+
|---|---|---|
|
| 15 |
+
| `equiv` | all-nli triplets | semantic equivalence |
|
| 16 |
+
| `simplify` | simple-wiki + altlex + sentence-compression | simplification / compression |
|
| 17 |
+
|
| 18 |
+
## Results
|
| 19 |
+
|
| 20 |
+
| config | STS-B rho | SICK-R rho |
|
| 21 |
+
|---|---|---|
|
| 22 |
+
| bare trunk | .5747 | .6526 |
|
| 23 |
+
| `equiv` alone | .7254 | **.7550** |
|
| 24 |
+
| `simplify` alone | .7400 | .7075 |
|
| 25 |
+
| **2-anchor dispatch** | **.7524** | .7380 |
|
| 26 |
+
|
| 27 |
+
The two are complementary along the TASK axis -- `simplify` wins STS-B solo,
|
| 28 |
+
`equiv` wins SICK-R -- which is the precondition a mixture needs. SICK-R is
|
| 29 |
+
never trained on and is the honest transfer read.
|
| 30 |
+
|
| 31 |
+
## Why the dispatch works here
|
| 32 |
+
|
| 33 |
+
Dispatched amplitude is `(w_k/z) * sigmoid(gate_k) * consume_k(x)`, where
|
| 34 |
+
`w_k/z = sinh(u_k) / SUM_j cosh(u_j)` over ALL anchors (the damping law).
|
| 35 |
+
That gives ~1.0 only when one anchor engages and the other **abstains**
|
| 36 |
+
(`u ~ 0`); if both fire it collapses to ~0.5 and the mixture delivers HALF of
|
| 37 |
+
what either member does alone.
|
| 38 |
+
|
| 39 |
+
Measured here: mean `|w/z|` moved from **.310/.380 (blend)** before alignment to
|
| 40 |
+
**.645/.223 (specialize)** after 800 keys-only steps, with no starvation
|
| 41 |
+
strikes. That flip is why the mixture beats its best member rather than damping
|
| 42 |
+
itself below it.
|
| 43 |
+
|
| 44 |
+
## Load
|
| 45 |
+
|
| 46 |
+
```python
|
| 47 |
+
import amoe
|
| 48 |
+
h = amoe.attach(trunk, ["amoe/moe/equiv.anchor.pt", "amoe/moe/simplify.anchor.pt"],
|
| 49 |
+
dispatch="amoe/moe/captionbert-v2-moe.dispatch.pt",
|
| 50 |
+
binding=CaptionBertV2Binding(d=512)) # from modeling_captionbert.py
|
| 51 |
+
base = h.detach() # bit-exact or raises
|
| 52 |
+
```
|
| 53 |
+
|
| 54 |
+
All anchors disabled reproduces the bare trunk **bit-exact** (asserted at build
|
| 55 |
+
time). Masking never renormalizes -- that is the damping law, not an oversight.
|
| 56 |
+
|
| 57 |
+
Anchor `.pt` key layout is `{block}.{param}`. (`blocks.{site}.{param}` in the
|
| 58 |
+
amoe README is the *safetensors* layout, a different serializer.)
|
| 59 |
+
|
| 60 |
+
## Training
|
| 61 |
+
|
| 62 |
+
Anchors: MNRL, in-batch + hard negatives where the source has them, 1,500 steps
|
| 63 |
+
at batch 256, pure Adam wd=0 (`amoe.laws.make_optimizer`), fp32/TF32 off.
|
| 64 |
+
1,500 not 4,000: the first solo run's STS-B **peaked at step 1,000** and then
|
| 65 |
+
fell .0178 while SICK-R kept climbing -- 56,825 distinct anchors behind 200,000
|
| 66 |
+
draws (space/draws .284). Dispatch: 800 steps, routing keys only (1,536 params),
|
| 67 |
+
anchors frozen, starvation safeguard armed.
|
| 68 |
+
|
| 69 |
+
See `metrics.json` for the full table and the routing telemetry.
|
amoe/collective/captionbert-v2-collective.dispatch.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:297141b94428df249f37ab29f0b40e63bd308026ac59400d319f4456d0e6c4c0
|
| 3 |
+
size 1590661
|
amoe/collective/config.json
ADDED
|
@@ -0,0 +1,153 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"run_name": "captionbert-v2-collective",
|
| 3 |
+
"trunk_repo": "AbstractPhil/captionbert-8192-v2",
|
| 4 |
+
"trunk_ckpt": "checkpoints/best_model.pt",
|
| 5 |
+
"tokenizer": "google-bert/bert-base-uncased",
|
| 6 |
+
"d_model": 512,
|
| 7 |
+
"n_heads": 8,
|
| 8 |
+
"n_layers": 12,
|
| 9 |
+
"d_ff": 2048,
|
| 10 |
+
"max_len": 8192,
|
| 11 |
+
"output_dim": 768,
|
| 12 |
+
"pooling": "mean",
|
| 13 |
+
"arms": [
|
| 14 |
+
[
|
| 15 |
+
"equiv",
|
| 16 |
+
{
|
| 17 |
+
"kind": "hub",
|
| 18 |
+
"path": "amoe/moe/equiv.anchor.pt",
|
| 19 |
+
"sources": [
|
| 20 |
+
[
|
| 21 |
+
"sentence-transformers/all-nli",
|
| 22 |
+
"triplet",
|
| 23 |
+
"anchor",
|
| 24 |
+
"positive",
|
| 25 |
+
"negative",
|
| 26 |
+
200000
|
| 27 |
+
]
|
| 28 |
+
]
|
| 29 |
+
}
|
| 30 |
+
],
|
| 31 |
+
[
|
| 32 |
+
"simplify",
|
| 33 |
+
{
|
| 34 |
+
"kind": "hub",
|
| 35 |
+
"path": "amoe/moe/simplify.anchor.pt",
|
| 36 |
+
"sources": [
|
| 37 |
+
[
|
| 38 |
+
"sentence-transformers/simple-wiki",
|
| 39 |
+
"pair",
|
| 40 |
+
"text",
|
| 41 |
+
"simplified",
|
| 42 |
+
null,
|
| 43 |
+
0
|
| 44 |
+
],
|
| 45 |
+
[
|
| 46 |
+
"sentence-transformers/altlex",
|
| 47 |
+
"pair",
|
| 48 |
+
"text",
|
| 49 |
+
"simplified",
|
| 50 |
+
null,
|
| 51 |
+
0
|
| 52 |
+
],
|
| 53 |
+
[
|
| 54 |
+
"sentence-transformers/sentence-compression",
|
| 55 |
+
"pair",
|
| 56 |
+
"text",
|
| 57 |
+
"simplified",
|
| 58 |
+
null,
|
| 59 |
+
0
|
| 60 |
+
]
|
| 61 |
+
]
|
| 62 |
+
}
|
| 63 |
+
],
|
| 64 |
+
[
|
| 65 |
+
"paraphrase",
|
| 66 |
+
{
|
| 67 |
+
"kind": "data",
|
| 68 |
+
"sources": [
|
| 69 |
+
[
|
| 70 |
+
"sentence-transformers/quora-duplicates",
|
| 71 |
+
"triplet",
|
| 72 |
+
"anchor",
|
| 73 |
+
"positive",
|
| 74 |
+
"negative",
|
| 75 |
+
0
|
| 76 |
+
]
|
| 77 |
+
]
|
| 78 |
+
}
|
| 79 |
+
],
|
| 80 |
+
[
|
| 81 |
+
"lexical",
|
| 82 |
+
{
|
| 83 |
+
"kind": "wordnet"
|
| 84 |
+
}
|
| 85 |
+
],
|
| 86 |
+
[
|
| 87 |
+
"topical",
|
| 88 |
+
{
|
| 89 |
+
"kind": "data",
|
| 90 |
+
"sources": [
|
| 91 |
+
[
|
| 92 |
+
"sentence-transformers/specter",
|
| 93 |
+
"triplet",
|
| 94 |
+
"anchor",
|
| 95 |
+
"positive",
|
| 96 |
+
"negative",
|
| 97 |
+
200000
|
| 98 |
+
]
|
| 99 |
+
]
|
| 100 |
+
}
|
| 101 |
+
],
|
| 102 |
+
[
|
| 103 |
+
"random",
|
| 104 |
+
{
|
| 105 |
+
"kind": "random"
|
| 106 |
+
}
|
| 107 |
+
]
|
| 108 |
+
],
|
| 109 |
+
"wordnet_max": 156000,
|
| 110 |
+
"dedup_jaccard": 0.95,
|
| 111 |
+
"base_arms": [
|
| 112 |
+
"equiv",
|
| 113 |
+
"simplify"
|
| 114 |
+
],
|
| 115 |
+
"greedy": true,
|
| 116 |
+
"greedy_rounds": 4,
|
| 117 |
+
"select_on": "STS-B",
|
| 118 |
+
"n_slots": 16,
|
| 119 |
+
"K": 64,
|
| 120 |
+
"D": 4,
|
| 121 |
+
"tau": 0.1,
|
| 122 |
+
"hidden": 178,
|
| 123 |
+
"gate_init": -3.0,
|
| 124 |
+
"anchor_steps": 1500,
|
| 125 |
+
"anchor_lr": 0.001,
|
| 126 |
+
"batch_size": 256,
|
| 127 |
+
"temperature": 0.05,
|
| 128 |
+
"max_tokens": 64,
|
| 129 |
+
"align_steps": 800,
|
| 130 |
+
"align_lr": 0.001,
|
| 131 |
+
"align_emb": 64,
|
| 132 |
+
"check_every": 200,
|
| 133 |
+
"usage_ppl_floor": 1.5,
|
| 134 |
+
"usage_min": 0.02,
|
| 135 |
+
"max_strikes": 3,
|
| 136 |
+
"mix_rows": true,
|
| 137 |
+
"align_taus": [
|
| 138 |
+
0.1
|
| 139 |
+
],
|
| 140 |
+
"align_seeds": [
|
| 141 |
+
0
|
| 142 |
+
],
|
| 143 |
+
"per_block_report": true,
|
| 144 |
+
"seed": 0,
|
| 145 |
+
"log_every": 100,
|
| 146 |
+
"eval_every": 500,
|
| 147 |
+
"reuse_anchors": true,
|
| 148 |
+
"out_dir": "/content/amoe_collective",
|
| 149 |
+
"hf_repo": "AbstractPhil/captionbert-8192-v2",
|
| 150 |
+
"hf_path": "amoe/collective",
|
| 151 |
+
"hf_private": false,
|
| 152 |
+
"hf_push": true
|
| 153 |
+
}
|
amoe/collective/metrics.json
ADDED
|
@@ -0,0 +1,1048 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"baseline": {
|
| 3 |
+
"STS-B": {
|
| 4 |
+
"spearman": 0.574721280584061
|
| 5 |
+
},
|
| 6 |
+
"SICK-R": {
|
| 7 |
+
"spearman": 0.652602297708298
|
| 8 |
+
},
|
| 9 |
+
"STS12": {
|
| 10 |
+
"spearman": 0.5051005991952476
|
| 11 |
+
},
|
| 12 |
+
"STS13": {
|
| 13 |
+
"spearman": 0.5994828037033634
|
| 14 |
+
},
|
| 15 |
+
"STS14": {
|
| 16 |
+
"spearman": 0.5451919606477016
|
| 17 |
+
},
|
| 18 |
+
"STS15": {
|
| 19 |
+
"spearman": 0.7136496565205896
|
| 20 |
+
},
|
| 21 |
+
"STS16": {
|
| 22 |
+
"spearman": 0.6776026025930126
|
| 23 |
+
},
|
| 24 |
+
"BIOSSES": {
|
| 25 |
+
"spearman": 0.5932785784958415
|
| 26 |
+
}
|
| 27 |
+
},
|
| 28 |
+
"rows": {
|
| 29 |
+
"OFF": {
|
| 30 |
+
"STS-B": {
|
| 31 |
+
"spearman": 0.574721280584061,
|
| 32 |
+
"self_cos": 0.13962045311927795,
|
| 33 |
+
"erank": 36.611595622425114
|
| 34 |
+
},
|
| 35 |
+
"SICK-R": {
|
| 36 |
+
"spearman": 0.652602297708298,
|
| 37 |
+
"self_cos": 0.32337328791618347,
|
| 38 |
+
"erank": 39.12520123844512
|
| 39 |
+
},
|
| 40 |
+
"STS12": {
|
| 41 |
+
"spearman": 0.5051005991952476,
|
| 42 |
+
"self_cos": 0.15351814031600952,
|
| 43 |
+
"erank": 25.252985927333295
|
| 44 |
+
},
|
| 45 |
+
"STS13": {
|
| 46 |
+
"spearman": 0.5994828037033634,
|
| 47 |
+
"self_cos": 0.15675018727779388,
|
| 48 |
+
"erank": 21.99929661499866
|
| 49 |
+
},
|
| 50 |
+
"STS14": {
|
| 51 |
+
"spearman": 0.5451919606477016,
|
| 52 |
+
"self_cos": 0.1182328462600708,
|
| 53 |
+
"erank": 36.69086900214419
|
| 54 |
+
},
|
| 55 |
+
"STS15": {
|
| 56 |
+
"spearman": 0.7136496565205896,
|
| 57 |
+
"self_cos": 0.1551932394504547,
|
| 58 |
+
"erank": 28.50109631029655
|
| 59 |
+
},
|
| 60 |
+
"STS16": {
|
| 61 |
+
"spearman": 0.6776026025930126,
|
| 62 |
+
"self_cos": 0.15450556576251984,
|
| 63 |
+
"erank": 31.893961189106346
|
| 64 |
+
},
|
| 65 |
+
"BIOSSES": {
|
| 66 |
+
"spearman": 0.5932785784958415,
|
| 67 |
+
"self_cos": 0.5799180269241333,
|
| 68 |
+
"erank": 24.42789597750344
|
| 69 |
+
}
|
| 70 |
+
},
|
| 71 |
+
"equiv-only": {
|
| 72 |
+
"STS-B": {
|
| 73 |
+
"spearman": 0.7208054927143117,
|
| 74 |
+
"self_cos": 0.11197277903556824,
|
| 75 |
+
"erank": 51.30250361724497
|
| 76 |
+
},
|
| 77 |
+
"SICK-R": {
|
| 78 |
+
"spearman": 0.7269011204456198,
|
| 79 |
+
"self_cos": 0.17548967897891998,
|
| 80 |
+
"erank": 35.771535966993405
|
| 81 |
+
},
|
| 82 |
+
"STS12": {
|
| 83 |
+
"spearman": 0.627535337359207,
|
| 84 |
+
"self_cos": 0.1346980780363083,
|
| 85 |
+
"erank": 43.12544739377829
|
| 86 |
+
},
|
| 87 |
+
"STS13": {
|
| 88 |
+
"spearman": 0.6988259672567189,
|
| 89 |
+
"self_cos": 0.18279729783535004,
|
| 90 |
+
"erank": 26.392917765960405
|
| 91 |
+
},
|
| 92 |
+
"STS14": {
|
| 93 |
+
"spearman": 0.6423465381918999,
|
| 94 |
+
"self_cos": 0.13812822103500366,
|
| 95 |
+
"erank": 44.65619024693345
|
| 96 |
+
},
|
| 97 |
+
"STS15": {
|
| 98 |
+
"spearman": 0.778300316950493,
|
| 99 |
+
"self_cos": 0.17851607501506805,
|
| 100 |
+
"erank": 29.35145982672314
|
| 101 |
+
},
|
| 102 |
+
"STS16": {
|
| 103 |
+
"spearman": 0.7107364005985578,
|
| 104 |
+
"self_cos": 0.1693350076675415,
|
| 105 |
+
"erank": 38.183249320162716
|
| 106 |
+
},
|
| 107 |
+
"BIOSSES": {
|
| 108 |
+
"spearman": 0.5845313912334387,
|
| 109 |
+
"self_cos": 0.5771422386169434,
|
| 110 |
+
"erank": 24.779312332397197
|
| 111 |
+
}
|
| 112 |
+
},
|
| 113 |
+
"simplify-only": {
|
| 114 |
+
"STS-B": {
|
| 115 |
+
"spearman": 0.6022308697258564,
|
| 116 |
+
"self_cos": 0.12909774482250214,
|
| 117 |
+
"erank": 40.86872085143915
|
| 118 |
+
},
|
| 119 |
+
"SICK-R": {
|
| 120 |
+
"spearman": 0.6579622689175231,
|
| 121 |
+
"self_cos": 0.31778937578201294,
|
| 122 |
+
"erank": 39.1674860276041
|
| 123 |
+
},
|
| 124 |
+
"STS12": {
|
| 125 |
+
"spearman": 0.5358899241682491,
|
| 126 |
+
"self_cos": 0.12675561010837555,
|
| 127 |
+
"erank": 29.405182004775746
|
| 128 |
+
},
|
| 129 |
+
"STS13": {
|
| 130 |
+
"spearman": 0.6304070306638541,
|
| 131 |
+
"self_cos": 0.12737733125686646,
|
| 132 |
+
"erank": 25.773887861093588
|
| 133 |
+
},
|
| 134 |
+
"STS14": {
|
| 135 |
+
"spearman": 0.574713371391879,
|
| 136 |
+
"self_cos": 0.0941900983452797,
|
| 137 |
+
"erank": 44.49932641739625
|
| 138 |
+
},
|
| 139 |
+
"STS15": {
|
| 140 |
+
"spearman": 0.7390672576622825,
|
| 141 |
+
"self_cos": 0.13394802808761597,
|
| 142 |
+
"erank": 31.20321830851873
|
| 143 |
+
},
|
| 144 |
+
"STS16": {
|
| 145 |
+
"spearman": 0.6945216097942847,
|
| 146 |
+
"self_cos": 0.1314028948545456,
|
| 147 |
+
"erank": 37.01869988293351
|
| 148 |
+
},
|
| 149 |
+
"BIOSSES": {
|
| 150 |
+
"spearman": 0.6015799724680118,
|
| 151 |
+
"self_cos": 0.543942928314209,
|
| 152 |
+
"erank": 26.957094005967072
|
| 153 |
+
}
|
| 154 |
+
},
|
| 155 |
+
"paraphrase-only": {
|
| 156 |
+
"STS-B": {
|
| 157 |
+
"spearman": 0.6126504112146641,
|
| 158 |
+
"self_cos": 0.1236109659075737,
|
| 159 |
+
"erank": 40.27058086660668
|
| 160 |
+
},
|
| 161 |
+
"SICK-R": {
|
| 162 |
+
"spearman": 0.6585007915483847,
|
| 163 |
+
"self_cos": 0.3122204840183258,
|
| 164 |
+
"erank": 39.76888407699623
|
| 165 |
+
},
|
| 166 |
+
"STS12": {
|
| 167 |
+
"spearman": 0.5154376837000663,
|
| 168 |
+
"self_cos": 0.1281055063009262,
|
| 169 |
+
"erank": 27.686515102668366
|
| 170 |
+
},
|
| 171 |
+
"STS13": {
|
| 172 |
+
"spearman": 0.6337825082750937,
|
| 173 |
+
"self_cos": 0.13173335790634155,
|
| 174 |
+
"erank": 24.82290178054701
|
| 175 |
+
},
|
| 176 |
+
"STS14": {
|
| 177 |
+
"spearman": 0.5773936947163352,
|
| 178 |
+
"self_cos": 0.0926285982131958,
|
| 179 |
+
"erank": 41.829968090608325
|
| 180 |
+
},
|
| 181 |
+
"STS15": {
|
| 182 |
+
"spearman": 0.7328088028392022,
|
| 183 |
+
"self_cos": 0.11987972259521484,
|
| 184 |
+
"erank": 31.822305225101996
|
| 185 |
+
},
|
| 186 |
+
"STS16": {
|
| 187 |
+
"spearman": 0.7331701300054091,
|
| 188 |
+
"self_cos": 0.10515344887971878,
|
| 189 |
+
"erank": 39.51330118651005
|
| 190 |
+
},
|
| 191 |
+
"BIOSSES": {
|
| 192 |
+
"spearman": 0.6051282460889176,
|
| 193 |
+
"self_cos": 0.5481609106063843,
|
| 194 |
+
"erank": 26.474269892906314
|
| 195 |
+
}
|
| 196 |
+
},
|
| 197 |
+
"COLLECTIVE": {
|
| 198 |
+
"STS-B": {
|
| 199 |
+
"spearman": 0.7683668046957611,
|
| 200 |
+
"self_cos": 0.09640415012836456,
|
| 201 |
+
"erank": 57.57577576807335
|
| 202 |
+
},
|
| 203 |
+
"SICK-R": {
|
| 204 |
+
"spearman": 0.7391220566292676,
|
| 205 |
+
"self_cos": 0.15953481197357178,
|
| 206 |
+
"erank": 35.99198048981848
|
| 207 |
+
},
|
| 208 |
+
"STS12": {
|
| 209 |
+
"spearman": 0.6682322345244319,
|
| 210 |
+
"self_cos": 0.105603888630867,
|
| 211 |
+
"erank": 53.024332799692715
|
| 212 |
+
},
|
| 213 |
+
"STS13": {
|
| 214 |
+
"spearman": 0.7557320871375747,
|
| 215 |
+
"self_cos": 0.1316331923007965,
|
| 216 |
+
"erank": 36.46135418608761
|
| 217 |
+
},
|
| 218 |
+
"STS14": {
|
| 219 |
+
"spearman": 0.6920853368670368,
|
| 220 |
+
"self_cos": 0.09459435939788818,
|
| 221 |
+
"erank": 58.71065398215481
|
| 222 |
+
},
|
| 223 |
+
"STS15": {
|
| 224 |
+
"spearman": 0.8055463823891014,
|
| 225 |
+
"self_cos": 0.13524731993675232,
|
| 226 |
+
"erank": 34.37401474216382
|
| 227 |
+
},
|
| 228 |
+
"STS16": {
|
| 229 |
+
"spearman": 0.7626482936903254,
|
| 230 |
+
"self_cos": 0.12244462966918945,
|
| 231 |
+
"erank": 53.00405416178429
|
| 232 |
+
},
|
| 233 |
+
"BIOSSES": {
|
| 234 |
+
"spearman": 0.6382494827064735,
|
| 235 |
+
"self_cos": 0.4860020875930786,
|
| 236 |
+
"erank": 30.829771511104333
|
| 237 |
+
}
|
| 238 |
+
}
|
| 239 |
+
},
|
| 240 |
+
"telemetry_before": {
|
| 241 |
+
"equiv": 0.23744726181030273,
|
| 242 |
+
"simplify": 0.22890318930149078,
|
| 243 |
+
"paraphrase": 0.2633548974990845
|
| 244 |
+
},
|
| 245 |
+
"telemetry_after": {
|
| 246 |
+
"equiv": 0.5873097777366638,
|
| 247 |
+
"simplify": 0.1378592997789383,
|
| 248 |
+
"paraphrase": 0.1424914002418518
|
| 249 |
+
},
|
| 250 |
+
"alarms": [],
|
| 251 |
+
"sweep": [
|
| 252 |
+
{
|
| 253 |
+
"subset": [
|
| 254 |
+
"equiv",
|
| 255 |
+
"simplify"
|
| 256 |
+
],
|
| 257 |
+
"tau": 0.1,
|
| 258 |
+
"seed": 0,
|
| 259 |
+
"tel_before": {
|
| 260 |
+
"equiv": 0.3686778247356415,
|
| 261 |
+
"simplify": 0.3205741345882416
|
| 262 |
+
},
|
| 263 |
+
"tel_after": {
|
| 264 |
+
"equiv": 0.6838898658752441,
|
| 265 |
+
"simplify": 0.19119425117969513
|
| 266 |
+
},
|
| 267 |
+
"full": {
|
| 268 |
+
"STS-B": {
|
| 269 |
+
"spearman": 0.7492622354327121,
|
| 270 |
+
"self_cos": 0.11812007427215576,
|
| 271 |
+
"erank": 53.50937322735125
|
| 272 |
+
},
|
| 273 |
+
"SICK-R": {
|
| 274 |
+
"spearman": 0.73649969707021,
|
| 275 |
+
"self_cos": 0.15910400450229645,
|
| 276 |
+
"erank": 34.26450305417673
|
| 277 |
+
},
|
| 278 |
+
"STS12": {
|
| 279 |
+
"spearman": 0.6603199239615669,
|
| 280 |
+
"self_cos": 0.12920618057250977,
|
| 281 |
+
"erank": 50.950827975327066
|
| 282 |
+
},
|
| 283 |
+
"STS13": {
|
| 284 |
+
"spearman": 0.7412117623295272,
|
| 285 |
+
"self_cos": 0.16888578236103058,
|
| 286 |
+
"erank": 34.00218908177754
|
| 287 |
+
},
|
| 288 |
+
"STS14": {
|
| 289 |
+
"spearman": 0.6882923298970814,
|
| 290 |
+
"self_cos": 0.12499123811721802,
|
| 291 |
+
"erank": 56.17271619952451
|
| 292 |
+
},
|
| 293 |
+
"STS15": {
|
| 294 |
+
"spearman": 0.8039666482327434,
|
| 295 |
+
"self_cos": 0.18052753806114197,
|
| 296 |
+
"erank": 31.453784346969037
|
| 297 |
+
},
|
| 298 |
+
"STS16": {
|
| 299 |
+
"spearman": 0.7366740362061548,
|
| 300 |
+
"self_cos": 0.1803213208913803,
|
| 301 |
+
"erank": 46.64553604673407
|
| 302 |
+
},
|
| 303 |
+
"BIOSSES": {
|
| 304 |
+
"spearman": 0.6700152667523759,
|
| 305 |
+
"self_cos": 0.5107892155647278,
|
| 306 |
+
"erank": 30.70854983961714
|
| 307 |
+
}
|
| 308 |
+
},
|
| 309 |
+
"singles": {
|
| 310 |
+
"equiv": 0.726814043958624,
|
| 311 |
+
"simplify": 0.6086975640630304
|
| 312 |
+
},
|
| 313 |
+
"off": 0.574721280584061,
|
| 314 |
+
"alarms": [],
|
| 315 |
+
"score": 0.7492622354327121
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"subset": [
|
| 319 |
+
"equiv",
|
| 320 |
+
"simplify",
|
| 321 |
+
"paraphrase"
|
| 322 |
+
],
|
| 323 |
+
"tau": 0.1,
|
| 324 |
+
"seed": 0,
|
| 325 |
+
"tel_before": {
|
| 326 |
+
"equiv": 0.23744726181030273,
|
| 327 |
+
"simplify": 0.22890318930149078,
|
| 328 |
+
"paraphrase": 0.2633548974990845
|
| 329 |
+
},
|
| 330 |
+
"tel_after": {
|
| 331 |
+
"equiv": 0.5873097777366638,
|
| 332 |
+
"simplify": 0.1378592997789383,
|
| 333 |
+
"paraphrase": 0.1424914002418518
|
| 334 |
+
},
|
| 335 |
+
"full": {
|
| 336 |
+
"STS-B": {
|
| 337 |
+
"spearman": 0.7683668046957611,
|
| 338 |
+
"self_cos": 0.09640415012836456,
|
| 339 |
+
"erank": 57.57577576807335
|
| 340 |
+
},
|
| 341 |
+
"SICK-R": {
|
| 342 |
+
"spearman": 0.7391220566292676,
|
| 343 |
+
"self_cos": 0.15953481197357178,
|
| 344 |
+
"erank": 35.99198048981848
|
| 345 |
+
},
|
| 346 |
+
"STS12": {
|
| 347 |
+
"spearman": 0.6682322345244319,
|
| 348 |
+
"self_cos": 0.105603888630867,
|
| 349 |
+
"erank": 53.024332799692715
|
| 350 |
+
},
|
| 351 |
+
"STS13": {
|
| 352 |
+
"spearman": 0.7557320871375747,
|
| 353 |
+
"self_cos": 0.1316331923007965,
|
| 354 |
+
"erank": 36.46135418608761
|
| 355 |
+
},
|
| 356 |
+
"STS14": {
|
| 357 |
+
"spearman": 0.6920853368670368,
|
| 358 |
+
"self_cos": 0.09459435939788818,
|
| 359 |
+
"erank": 58.71065398215481
|
| 360 |
+
},
|
| 361 |
+
"STS15": {
|
| 362 |
+
"spearman": 0.8055463823891014,
|
| 363 |
+
"self_cos": 0.13524731993675232,
|
| 364 |
+
"erank": 34.37401474216382
|
| 365 |
+
},
|
| 366 |
+
"STS16": {
|
| 367 |
+
"spearman": 0.7626482936903254,
|
| 368 |
+
"self_cos": 0.12244462966918945,
|
| 369 |
+
"erank": 53.00405416178429
|
| 370 |
+
},
|
| 371 |
+
"BIOSSES": {
|
| 372 |
+
"spearman": 0.6382494827064735,
|
| 373 |
+
"self_cos": 0.4860020875930786,
|
| 374 |
+
"erank": 30.829771511104333
|
| 375 |
+
}
|
| 376 |
+
},
|
| 377 |
+
"singles": {
|
| 378 |
+
"equiv": 0.7208054927143117,
|
| 379 |
+
"simplify": 0.6022308697258564,
|
| 380 |
+
"paraphrase": 0.6126504112146641
|
| 381 |
+
},
|
| 382 |
+
"off": 0.574721280584061,
|
| 383 |
+
"alarms": [],
|
| 384 |
+
"score": 0.7683668046957611
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"subset": [
|
| 388 |
+
"equiv",
|
| 389 |
+
"simplify",
|
| 390 |
+
"lexical"
|
| 391 |
+
],
|
| 392 |
+
"tau": 0.1,
|
| 393 |
+
"seed": 0,
|
| 394 |
+
"tel_before": {
|
| 395 |
+
"equiv": 0.24023212492465973,
|
| 396 |
+
"simplify": 0.23160450160503387,
|
| 397 |
+
"lexical": 0.26030921936035156
|
| 398 |
+
},
|
| 399 |
+
"tel_after": {
|
| 400 |
+
"equiv": 0.5473803877830505,
|
| 401 |
+
"simplify": 0.12730351090431213,
|
| 402 |
+
"lexical": 0.19721834361553192
|
| 403 |
+
},
|
| 404 |
+
"full": {
|
| 405 |
+
"STS-B": {
|
| 406 |
+
"spearman": 0.7517867527708854,
|
| 407 |
+
"self_cos": 0.1069379523396492,
|
| 408 |
+
"erank": 53.515357929991715
|
| 409 |
+
},
|
| 410 |
+
"SICK-R": {
|
| 411 |
+
"spearman": 0.7384364269431155,
|
| 412 |
+
"self_cos": 0.17880664765834808,
|
| 413 |
+
"erank": 35.33930782439967
|
| 414 |
+
},
|
| 415 |
+
"STS12": {
|
| 416 |
+
"spearman": 0.6675660047117302,
|
| 417 |
+
"self_cos": 0.12142571806907654,
|
| 418 |
+
"erank": 48.20131169397737
|
| 419 |
+
},
|
| 420 |
+
"STS13": {
|
| 421 |
+
"spearman": 0.7463762525241542,
|
| 422 |
+
"self_cos": 0.1441446989774704,
|
| 423 |
+
"erank": 30.71307488039915
|
| 424 |
+
},
|
| 425 |
+
"STS14": {
|
| 426 |
+
"spearman": 0.6807728253135723,
|
| 427 |
+
"self_cos": 0.1154785007238388,
|
| 428 |
+
"erank": 52.230237809011335
|
| 429 |
+
},
|
| 430 |
+
"STS15": {
|
| 431 |
+
"spearman": 0.7992807220785836,
|
| 432 |
+
"self_cos": 0.17022141814231873,
|
| 433 |
+
"erank": 32.533574462111254
|
| 434 |
+
},
|
| 435 |
+
"STS16": {
|
| 436 |
+
"spearman": 0.7432641868699787,
|
| 437 |
+
"self_cos": 0.16130828857421875,
|
| 438 |
+
"erank": 43.947130064788446
|
| 439 |
+
},
|
| 440 |
+
"BIOSSES": {
|
| 441 |
+
"spearman": 0.5890917361078457,
|
| 442 |
+
"self_cos": 0.517375648021698,
|
| 443 |
+
"erank": 31.035833403472914
|
| 444 |
+
}
|
| 445 |
+
},
|
| 446 |
+
"singles": {
|
| 447 |
+
"equiv": 0.7134587810893606,
|
| 448 |
+
"simplify": 0.6085805955499919,
|
| 449 |
+
"lexical": 0.5848311265049561
|
| 450 |
+
},
|
| 451 |
+
"off": 0.574721280584061,
|
| 452 |
+
"alarms": [],
|
| 453 |
+
"score": 0.7517867527708854
|
| 454 |
+
},
|
| 455 |
+
{
|
| 456 |
+
"subset": [
|
| 457 |
+
"equiv",
|
| 458 |
+
"simplify",
|
| 459 |
+
"topical"
|
| 460 |
+
],
|
| 461 |
+
"tau": 0.1,
|
| 462 |
+
"seed": 0,
|
| 463 |
+
"tel_before": {
|
| 464 |
+
"equiv": 0.2420409917831421,
|
| 465 |
+
"simplify": 0.2299657016992569,
|
| 466 |
+
"topical": 0.25807517766952515
|
| 467 |
+
},
|
| 468 |
+
"tel_after": {
|
| 469 |
+
"equiv": 0.5332305431365967,
|
| 470 |
+
"simplify": 0.1281643956899643,
|
| 471 |
+
"topical": 0.22435426712036133
|
| 472 |
+
},
|
| 473 |
+
"full": {
|
| 474 |
+
"STS-B": {
|
| 475 |
+
"spearman": 0.7507712625705492,
|
| 476 |
+
"self_cos": 0.12828144431114197,
|
| 477 |
+
"erank": 52.96040953420221
|
| 478 |
+
},
|
| 479 |
+
"SICK-R": {
|
| 480 |
+
"spearman": 0.7359322363735815,
|
| 481 |
+
"self_cos": 0.17767615616321564,
|
| 482 |
+
"erank": 35.26025986619767
|
| 483 |
+
},
|
| 484 |
+
"STS12": {
|
| 485 |
+
"spearman": 0.6818500432458242,
|
| 486 |
+
"self_cos": 0.1493045538663864,
|
| 487 |
+
"erank": 48.48233990906185
|
| 488 |
+
},
|
| 489 |
+
"STS13": {
|
| 490 |
+
"spearman": 0.7483124044859479,
|
| 491 |
+
"self_cos": 0.20377053320407867,
|
| 492 |
+
"erank": 31.84735659964571
|
| 493 |
+
},
|
| 494 |
+
"STS14": {
|
| 495 |
+
"spearman": 0.6860237299421542,
|
| 496 |
+
"self_cos": 0.1415916234254837,
|
| 497 |
+
"erank": 53.587674150977946
|
| 498 |
+
},
|
| 499 |
+
"STS15": {
|
| 500 |
+
"spearman": 0.7958472119231931,
|
| 501 |
+
"self_cos": 0.21095603704452515,
|
| 502 |
+
"erank": 29.604773264550584
|
| 503 |
+
},
|
| 504 |
+
"STS16": {
|
| 505 |
+
"spearman": 0.7443357011449099,
|
| 506 |
+
"self_cos": 0.2329452484846115,
|
| 507 |
+
"erank": 45.51218030263531
|
| 508 |
+
},
|
| 509 |
+
"BIOSSES": {
|
| 510 |
+
"spearman": 0.6352976082170951,
|
| 511 |
+
"self_cos": 0.48390987515449524,
|
| 512 |
+
"erank": 33.796202057295986
|
| 513 |
+
}
|
| 514 |
+
},
|
| 515 |
+
"singles": {
|
| 516 |
+
"equiv": 0.7184094524436696,
|
| 517 |
+
"simplify": 0.6029975263774512,
|
| 518 |
+
"topical": 0.5892148693630693
|
| 519 |
+
},
|
| 520 |
+
"off": 0.574721280584061,
|
| 521 |
+
"alarms": [],
|
| 522 |
+
"score": 0.7507712625705492
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"subset": [
|
| 526 |
+
"equiv",
|
| 527 |
+
"simplify",
|
| 528 |
+
"random"
|
| 529 |
+
],
|
| 530 |
+
"tau": 0.1,
|
| 531 |
+
"seed": 0,
|
| 532 |
+
"tel_before": {
|
| 533 |
+
"equiv": 0.23760630190372467,
|
| 534 |
+
"simplify": 0.2314901500940323,
|
| 535 |
+
"random": 0.2617114782333374
|
| 536 |
+
},
|
| 537 |
+
"tel_after": {
|
| 538 |
+
"equiv": 0.5893611907958984,
|
| 539 |
+
"simplify": 0.1493033617734909,
|
| 540 |
+
"random": 0.13993841409683228
|
| 541 |
+
},
|
| 542 |
+
"full": {
|
| 543 |
+
"STS-B": {
|
| 544 |
+
"spearman": 0.7538404708132564,
|
| 545 |
+
"self_cos": 0.11281727999448776,
|
| 546 |
+
"erank": 55.19763084195173
|
| 547 |
+
},
|
| 548 |
+
"SICK-R": {
|
| 549 |
+
"spearman": 0.7352546449629594,
|
| 550 |
+
"self_cos": 0.16733522713184357,
|
| 551 |
+
"erank": 35.03743528242395
|
| 552 |
+
},
|
| 553 |
+
"STS12": {
|
| 554 |
+
"spearman": 0.6645941232111496,
|
| 555 |
+
"self_cos": 0.12341051548719406,
|
| 556 |
+
"erank": 51.19679124124194
|
| 557 |
+
},
|
| 558 |
+
"STS13": {
|
| 559 |
+
"spearman": 0.7406425601390596,
|
| 560 |
+
"self_cos": 0.15680178999900818,
|
| 561 |
+
"erank": 33.302242729527805
|
| 562 |
+
},
|
| 563 |
+
"STS14": {
|
| 564 |
+
"spearman": 0.6810938554899827,
|
| 565 |
+
"self_cos": 0.11642584204673767,
|
| 566 |
+
"erank": 55.75402332689603
|
| 567 |
+
},
|
| 568 |
+
"STS15": {
|
| 569 |
+
"spearman": 0.8022667862635331,
|
| 570 |
+
"self_cos": 0.17066547274589539,
|
| 571 |
+
"erank": 32.258335926066046
|
| 572 |
+
},
|
| 573 |
+
"STS16": {
|
| 574 |
+
"spearman": 0.7396458776164397,
|
| 575 |
+
"self_cos": 0.1694430112838745,
|
| 576 |
+
"erank": 46.72131715480854
|
| 577 |
+
},
|
| 578 |
+
"BIOSSES": {
|
| 579 |
+
"spearman": 0.6085680835040913,
|
| 580 |
+
"self_cos": 0.5228798389434814,
|
| 581 |
+
"erank": 28.50197360668008
|
| 582 |
+
}
|
| 583 |
+
},
|
| 584 |
+
"singles": {
|
| 585 |
+
"equiv": 0.7281240331204828,
|
| 586 |
+
"simplify": 0.6083812574902878,
|
| 587 |
+
"random": 0.5747175054849759
|
| 588 |
+
},
|
| 589 |
+
"off": 0.574721280584061,
|
| 590 |
+
"alarms": [],
|
| 591 |
+
"score": 0.7538404708132564
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"subset": [
|
| 595 |
+
"equiv",
|
| 596 |
+
"simplify",
|
| 597 |
+
"paraphrase",
|
| 598 |
+
"lexical"
|
| 599 |
+
],
|
| 600 |
+
"tau": 0.1,
|
| 601 |
+
"seed": 0,
|
| 602 |
+
"tel_before": {
|
| 603 |
+
"equiv": 0.16629481315612793,
|
| 604 |
+
"simplify": 0.16293124854564667,
|
| 605 |
+
"paraphrase": 0.205373153090477,
|
| 606 |
+
"lexical": 0.20466244220733643
|
| 607 |
+
},
|
| 608 |
+
"tel_after": {
|
| 609 |
+
"equiv": 0.46579989790916443,
|
| 610 |
+
"simplify": 0.12858204543590546,
|
| 611 |
+
"paraphrase": 0.1340203434228897,
|
| 612 |
+
"lexical": 0.16869686543941498
|
| 613 |
+
},
|
| 614 |
+
"full": {
|
| 615 |
+
"STS-B": {
|
| 616 |
+
"spearman": 0.7497524760306598,
|
| 617 |
+
"self_cos": 0.09558529406785965,
|
| 618 |
+
"erank": 54.721844231903276
|
| 619 |
+
},
|
| 620 |
+
"SICK-R": {
|
| 621 |
+
"spearman": 0.7279983472302813,
|
| 622 |
+
"self_cos": 0.1752961128950119,
|
| 623 |
+
"erank": 36.66820925835615
|
| 624 |
+
},
|
| 625 |
+
"STS12": {
|
| 626 |
+
"spearman": 0.6702679071576564,
|
| 627 |
+
"self_cos": 0.11095768213272095,
|
| 628 |
+
"erank": 47.884028137779445
|
| 629 |
+
},
|
| 630 |
+
"STS13": {
|
| 631 |
+
"spearman": 0.7487306187629342,
|
| 632 |
+
"self_cos": 0.1418977528810501,
|
| 633 |
+
"erank": 31.538345829404616
|
| 634 |
+
},
|
| 635 |
+
"STS14": {
|
| 636 |
+
"spearman": 0.6855877208080531,
|
| 637 |
+
"self_cos": 0.10578248649835587,
|
| 638 |
+
"erank": 51.3322767094024
|
| 639 |
+
},
|
| 640 |
+
"STS15": {
|
| 641 |
+
"spearman": 0.7967253337162304,
|
| 642 |
+
"self_cos": 0.14615561068058014,
|
| 643 |
+
"erank": 33.327804503834855
|
| 644 |
+
},
|
| 645 |
+
"STS16": {
|
| 646 |
+
"spearman": 0.7554023899926602,
|
| 647 |
+
"self_cos": 0.12975861132144928,
|
| 648 |
+
"erank": 46.311823211423516
|
| 649 |
+
},
|
| 650 |
+
"BIOSSES": {
|
| 651 |
+
"spearman": 0.604537871191042,
|
| 652 |
+
"self_cos": 0.4957801103591919,
|
| 653 |
+
"erank": 32.2624831581336
|
| 654 |
+
}
|
| 655 |
+
},
|
| 656 |
+
"singles": {
|
| 657 |
+
"equiv": 0.6973822215513934,
|
| 658 |
+
"simplify": 0.603279233983821,
|
| 659 |
+
"paraphrase": 0.6034835680939526,
|
| 660 |
+
"lexical": 0.5816887986246279
|
| 661 |
+
},
|
| 662 |
+
"off": 0.574721280584061,
|
| 663 |
+
"alarms": [],
|
| 664 |
+
"score": 0.7497524760306598
|
| 665 |
+
},
|
| 666 |
+
{
|
| 667 |
+
"subset": [
|
| 668 |
+
"equiv",
|
| 669 |
+
"simplify",
|
| 670 |
+
"paraphrase",
|
| 671 |
+
"topical"
|
| 672 |
+
],
|
| 673 |
+
"tau": 0.1,
|
| 674 |
+
"seed": 0,
|
| 675 |
+
"tel_before": {
|
| 676 |
+
"equiv": 0.16726410388946533,
|
| 677 |
+
"simplify": 0.16440722346305847,
|
| 678 |
+
"paraphrase": 0.20589487254619598,
|
| 679 |
+
"topical": 0.20155394077301025
|
| 680 |
+
},
|
| 681 |
+
"tel_after": {
|
| 682 |
+
"equiv": 0.4615342617034912,
|
| 683 |
+
"simplify": 0.11386198550462723,
|
| 684 |
+
"paraphrase": 0.11478659510612488,
|
| 685 |
+
"topical": 0.2129145860671997
|
| 686 |
+
},
|
| 687 |
+
"full": {
|
| 688 |
+
"STS-B": {
|
| 689 |
+
"spearman": 0.7539168226503975,
|
| 690 |
+
"self_cos": 0.10456515848636627,
|
| 691 |
+
"erank": 55.70978505018554
|
| 692 |
+
},
|
| 693 |
+
"SICK-R": {
|
| 694 |
+
"spearman": 0.7278035145346904,
|
| 695 |
+
"self_cos": 0.17203262448310852,
|
| 696 |
+
"erank": 36.49386507010437
|
| 697 |
+
},
|
| 698 |
+
"STS12": {
|
| 699 |
+
"spearman": 0.6853824758006748,
|
| 700 |
+
"self_cos": 0.12454281002283096,
|
| 701 |
+
"erank": 48.698947468258204
|
| 702 |
+
},
|
| 703 |
+
"STS13": {
|
| 704 |
+
"spearman": 0.758641035446806,
|
| 705 |
+
"self_cos": 0.18651849031448364,
|
| 706 |
+
"erank": 34.40391320299821
|
| 707 |
+
},
|
| 708 |
+
"STS14": {
|
| 709 |
+
"spearman": 0.6960840850169944,
|
| 710 |
+
"self_cos": 0.11937179416418076,
|
| 711 |
+
"erank": 54.9126651501131
|
| 712 |
+
},
|
| 713 |
+
"STS15": {
|
| 714 |
+
"spearman": 0.8008059312937448,
|
| 715 |
+
"self_cos": 0.17032131552696228,
|
| 716 |
+
"erank": 31.791153994022174
|
| 717 |
+
},
|
| 718 |
+
"STS16": {
|
| 719 |
+
"spearman": 0.760447994939318,
|
| 720 |
+
"self_cos": 0.1732800453901291,
|
| 721 |
+
"erank": 49.38432238232354
|
| 722 |
+
},
|
| 723 |
+
"BIOSSES": {
|
| 724 |
+
"spearman": 0.6485689949519943,
|
| 725 |
+
"self_cos": 0.4674919545650482,
|
| 726 |
+
"erank": 34.8354198561506
|
| 727 |
+
}
|
| 728 |
+
},
|
| 729 |
+
"singles": {
|
| 730 |
+
"equiv": 0.6986657437867774,
|
| 731 |
+
"simplify": 0.5964483914814321,
|
| 732 |
+
"paraphrase": 0.6062877126170938,
|
| 733 |
+
"topical": 0.5852188605460859
|
| 734 |
+
},
|
| 735 |
+
"off": 0.574721280584061,
|
| 736 |
+
"alarms": [],
|
| 737 |
+
"score": 0.7539168226503975
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"subset": [
|
| 741 |
+
"equiv",
|
| 742 |
+
"simplify",
|
| 743 |
+
"paraphrase",
|
| 744 |
+
"random"
|
| 745 |
+
],
|
| 746 |
+
"tau": 0.1,
|
| 747 |
+
"seed": 0,
|
| 748 |
+
"tel_before": {
|
| 749 |
+
"equiv": 0.16712957620620728,
|
| 750 |
+
"simplify": 0.16495832800865173,
|
| 751 |
+
"paraphrase": 0.20430995523929596,
|
| 752 |
+
"random": 0.19984109699726105
|
| 753 |
+
},
|
| 754 |
+
"tel_after": {
|
| 755 |
+
"equiv": 0.5039264559745789,
|
| 756 |
+
"simplify": 0.1547492891550064,
|
| 757 |
+
"paraphrase": 0.13915561139583588,
|
| 758 |
+
"random": 0.08529075980186462
|
| 759 |
+
},
|
| 760 |
+
"full": {
|
| 761 |
+
"STS-B": {
|
| 762 |
+
"spearman": 0.7547194572790746,
|
| 763 |
+
"self_cos": 0.09596192091703415,
|
| 764 |
+
"erank": 57.03773415446015
|
| 765 |
+
},
|
| 766 |
+
"SICK-R": {
|
| 767 |
+
"spearman": 0.7281140523796689,
|
| 768 |
+
"self_cos": 0.16770263016223907,
|
| 769 |
+
"erank": 36.30635018861
|
| 770 |
+
},
|
| 771 |
+
"STS12": {
|
| 772 |
+
"spearman": 0.6717131780391491,
|
| 773 |
+
"self_cos": 0.10657699406147003,
|
| 774 |
+
"erank": 50.91815201449891
|
| 775 |
+
},
|
| 776 |
+
"STS13": {
|
| 777 |
+
"spearman": 0.7462598158896672,
|
| 778 |
+
"self_cos": 0.1420094519853592,
|
| 779 |
+
"erank": 34.721579091614444
|
| 780 |
+
},
|
| 781 |
+
"STS14": {
|
| 782 |
+
"spearman": 0.6858800082214228,
|
| 783 |
+
"self_cos": 0.10232123732566833,
|
| 784 |
+
"erank": 56.5016249628113
|
| 785 |
+
},
|
| 786 |
+
"STS15": {
|
| 787 |
+
"spearman": 0.8003883166728946,
|
| 788 |
+
"self_cos": 0.1399666965007782,
|
| 789 |
+
"erank": 33.433214375154385
|
| 790 |
+
},
|
| 791 |
+
"STS16": {
|
| 792 |
+
"spearman": 0.7594716635390275,
|
| 793 |
+
"self_cos": 0.13230887055397034,
|
| 794 |
+
"erank": 50.35027733566746
|
| 795 |
+
},
|
| 796 |
+
"BIOSSES": {
|
| 797 |
+
"spearman": 0.6325927273074403,
|
| 798 |
+
"self_cos": 0.4877208173274994,
|
| 799 |
+
"erank": 30.796648830453222
|
| 800 |
+
}
|
| 801 |
+
},
|
| 802 |
+
"singles": {
|
| 803 |
+
"equiv": 0.7090303218487345,
|
| 804 |
+
"simplify": 0.6064690919193181,
|
| 805 |
+
"paraphrase": 0.6068871520729853,
|
| 806 |
+
"random": 0.5747158424255487
|
| 807 |
+
},
|
| 808 |
+
"off": 0.574721280584061,
|
| 809 |
+
"alarms": [],
|
| 810 |
+
"score": 0.7547194572790746
|
| 811 |
+
}
|
| 812 |
+
],
|
| 813 |
+
"best": {
|
| 814 |
+
"subset": [
|
| 815 |
+
"equiv",
|
| 816 |
+
"simplify",
|
| 817 |
+
"paraphrase"
|
| 818 |
+
],
|
| 819 |
+
"tau": 0.1,
|
| 820 |
+
"seed": 0
|
| 821 |
+
},
|
| 822 |
+
"per_block_routing": {
|
| 823 |
+
"per_block": [
|
| 824 |
+
[
|
| 825 |
+
0.5217128992080688,
|
| 826 |
+
0.15346358716487885,
|
| 827 |
+
0.12478164583444595
|
| 828 |
+
],
|
| 829 |
+
[
|
| 830 |
+
0.379960298538208,
|
| 831 |
+
0.12925627827644348,
|
| 832 |
+
0.2731368839740753
|
| 833 |
+
],
|
| 834 |
+
[
|
| 835 |
+
0.6872727870941162,
|
| 836 |
+
0.12642249464988708,
|
| 837 |
+
0.09564091265201569
|
| 838 |
+
],
|
| 839 |
+
[
|
| 840 |
+
0.5433812737464905,
|
| 841 |
+
0.19714927673339844,
|
| 842 |
+
0.16204111278057098
|
| 843 |
+
],
|
| 844 |
+
[
|
| 845 |
+
0.4013535678386688,
|
| 846 |
+
0.21368706226348877,
|
| 847 |
+
0.18829737603664398
|
| 848 |
+
],
|
| 849 |
+
[
|
| 850 |
+
0.5704016089439392,
|
| 851 |
+
0.1158270537853241,
|
| 852 |
+
0.18047505617141724
|
| 853 |
+
],
|
| 854 |
+
[
|
| 855 |
+
0.5290986895561218,
|
| 856 |
+
0.20661617815494537,
|
| 857 |
+
0.13872331380844116
|
| 858 |
+
],
|
| 859 |
+
[
|
| 860 |
+
0.5848025679588318,
|
| 861 |
+
0.09634139388799667,
|
| 862 |
+
0.2000529021024704
|
| 863 |
+
],
|
| 864 |
+
[
|
| 865 |
+
0.7276835441589355,
|
| 866 |
+
0.09993541985750198,
|
| 867 |
+
0.0673375055193901
|
| 868 |
+
],
|
| 869 |
+
[
|
| 870 |
+
0.7168727517127991,
|
| 871 |
+
0.09174758195877075,
|
| 872 |
+
0.09276697039604187
|
| 873 |
+
],
|
| 874 |
+
[
|
| 875 |
+
0.695261538028717,
|
| 876 |
+
0.10963375121355057,
|
| 877 |
+
0.08987089991569519
|
| 878 |
+
],
|
| 879 |
+
[
|
| 880 |
+
0.6899158358573914,
|
| 881 |
+
0.11423160880804062,
|
| 882 |
+
0.09677205979824066
|
| 883 |
+
]
|
| 884 |
+
],
|
| 885 |
+
"spread": 0.34772324562072754
|
| 886 |
+
},
|
| 887 |
+
"anchors": {
|
| 888 |
+
"equiv": "equiv.anchor.pt",
|
| 889 |
+
"simplify": "simplify.anchor.pt",
|
| 890 |
+
"paraphrase": "paraphrase.anchor.pt",
|
| 891 |
+
"lexical": "lexical.anchor.pt",
|
| 892 |
+
"topical": "topical.anchor.pt",
|
| 893 |
+
"random": "random.anchor.pt"
|
| 894 |
+
},
|
| 895 |
+
"config": {
|
| 896 |
+
"run_name": "captionbert-v2-collective",
|
| 897 |
+
"trunk_repo": "AbstractPhil/captionbert-8192-v2",
|
| 898 |
+
"trunk_ckpt": "checkpoints/best_model.pt",
|
| 899 |
+
"tokenizer": "google-bert/bert-base-uncased",
|
| 900 |
+
"d_model": 512,
|
| 901 |
+
"n_heads": 8,
|
| 902 |
+
"n_layers": 12,
|
| 903 |
+
"d_ff": 2048,
|
| 904 |
+
"max_len": 8192,
|
| 905 |
+
"output_dim": 768,
|
| 906 |
+
"pooling": "mean",
|
| 907 |
+
"arms": [
|
| 908 |
+
[
|
| 909 |
+
"equiv",
|
| 910 |
+
{
|
| 911 |
+
"kind": "hub",
|
| 912 |
+
"path": "amoe/moe/equiv.anchor.pt",
|
| 913 |
+
"sources": [
|
| 914 |
+
[
|
| 915 |
+
"sentence-transformers/all-nli",
|
| 916 |
+
"triplet",
|
| 917 |
+
"anchor",
|
| 918 |
+
"positive",
|
| 919 |
+
"negative",
|
| 920 |
+
200000
|
| 921 |
+
]
|
| 922 |
+
]
|
| 923 |
+
}
|
| 924 |
+
],
|
| 925 |
+
[
|
| 926 |
+
"simplify",
|
| 927 |
+
{
|
| 928 |
+
"kind": "hub",
|
| 929 |
+
"path": "amoe/moe/simplify.anchor.pt",
|
| 930 |
+
"sources": [
|
| 931 |
+
[
|
| 932 |
+
"sentence-transformers/simple-wiki",
|
| 933 |
+
"pair",
|
| 934 |
+
"text",
|
| 935 |
+
"simplified",
|
| 936 |
+
null,
|
| 937 |
+
0
|
| 938 |
+
],
|
| 939 |
+
[
|
| 940 |
+
"sentence-transformers/altlex",
|
| 941 |
+
"pair",
|
| 942 |
+
"text",
|
| 943 |
+
"simplified",
|
| 944 |
+
null,
|
| 945 |
+
0
|
| 946 |
+
],
|
| 947 |
+
[
|
| 948 |
+
"sentence-transformers/sentence-compression",
|
| 949 |
+
"pair",
|
| 950 |
+
"text",
|
| 951 |
+
"simplified",
|
| 952 |
+
null,
|
| 953 |
+
0
|
| 954 |
+
]
|
| 955 |
+
]
|
| 956 |
+
}
|
| 957 |
+
],
|
| 958 |
+
[
|
| 959 |
+
"paraphrase",
|
| 960 |
+
{
|
| 961 |
+
"kind": "data",
|
| 962 |
+
"sources": [
|
| 963 |
+
[
|
| 964 |
+
"sentence-transformers/quora-duplicates",
|
| 965 |
+
"triplet",
|
| 966 |
+
"anchor",
|
| 967 |
+
"positive",
|
| 968 |
+
"negative",
|
| 969 |
+
0
|
| 970 |
+
]
|
| 971 |
+
]
|
| 972 |
+
}
|
| 973 |
+
],
|
| 974 |
+
[
|
| 975 |
+
"lexical",
|
| 976 |
+
{
|
| 977 |
+
"kind": "wordnet"
|
| 978 |
+
}
|
| 979 |
+
],
|
| 980 |
+
[
|
| 981 |
+
"topical",
|
| 982 |
+
{
|
| 983 |
+
"kind": "data",
|
| 984 |
+
"sources": [
|
| 985 |
+
[
|
| 986 |
+
"sentence-transformers/specter",
|
| 987 |
+
"triplet",
|
| 988 |
+
"anchor",
|
| 989 |
+
"positive",
|
| 990 |
+
"negative",
|
| 991 |
+
200000
|
| 992 |
+
]
|
| 993 |
+
]
|
| 994 |
+
}
|
| 995 |
+
],
|
| 996 |
+
[
|
| 997 |
+
"random",
|
| 998 |
+
{
|
| 999 |
+
"kind": "random"
|
| 1000 |
+
}
|
| 1001 |
+
]
|
| 1002 |
+
],
|
| 1003 |
+
"wordnet_max": 156000,
|
| 1004 |
+
"dedup_jaccard": 0.95,
|
| 1005 |
+
"base_arms": [
|
| 1006 |
+
"equiv",
|
| 1007 |
+
"simplify"
|
| 1008 |
+
],
|
| 1009 |
+
"greedy": true,
|
| 1010 |
+
"greedy_rounds": 4,
|
| 1011 |
+
"select_on": "STS-B",
|
| 1012 |
+
"n_slots": 16,
|
| 1013 |
+
"K": 64,
|
| 1014 |
+
"D": 4,
|
| 1015 |
+
"tau": 0.1,
|
| 1016 |
+
"hidden": 178,
|
| 1017 |
+
"gate_init": -3.0,
|
| 1018 |
+
"anchor_steps": 1500,
|
| 1019 |
+
"anchor_lr": 0.001,
|
| 1020 |
+
"batch_size": 256,
|
| 1021 |
+
"temperature": 0.05,
|
| 1022 |
+
"max_tokens": 64,
|
| 1023 |
+
"align_steps": 800,
|
| 1024 |
+
"align_lr": 0.001,
|
| 1025 |
+
"align_emb": 64,
|
| 1026 |
+
"check_every": 200,
|
| 1027 |
+
"usage_ppl_floor": 1.5,
|
| 1028 |
+
"usage_min": 0.02,
|
| 1029 |
+
"max_strikes": 3,
|
| 1030 |
+
"mix_rows": true,
|
| 1031 |
+
"align_taus": [
|
| 1032 |
+
0.1
|
| 1033 |
+
],
|
| 1034 |
+
"align_seeds": [
|
| 1035 |
+
0
|
| 1036 |
+
],
|
| 1037 |
+
"per_block_report": true,
|
| 1038 |
+
"seed": 0,
|
| 1039 |
+
"log_every": 100,
|
| 1040 |
+
"eval_every": 500,
|
| 1041 |
+
"reuse_anchors": true,
|
| 1042 |
+
"out_dir": "/content/amoe_collective",
|
| 1043 |
+
"hf_repo": "AbstractPhil/captionbert-8192-v2",
|
| 1044 |
+
"hf_path": "amoe/collective",
|
| 1045 |
+
"hf_private": false,
|
| 1046 |
+
"hf_push": true
|
| 1047 |
+
}
|
| 1048 |
+
}
|