Publish ESS-AIST-81M preview GGUF quantizations
Browse files- .gitattributes +1 -34
- ESS-AIST-81M_q5_1.gguf +3 -0
- ESS-AIST-81M_q8_0.gguf +3 -0
- README.md +72 -0
- event_eval.json +266 -0
- export_metadata.json +49 -0
- manifest.json +10 -0
- parameter_breakdown.json +9 -0
- prefix_eval.json +48 -0
- retrieval_512_gt1030.json +40 -0
- subject_eval.json +119 -0
.gitattributes
CHANGED
|
@@ -1,35 +1,2 @@
|
|
| 1 |
-
*.7z filter=lfs diff=lfs merge=lfs -text
|
| 2 |
-
*.arrow filter=lfs diff=lfs merge=lfs -text
|
| 3 |
-
*.bin filter=lfs diff=lfs merge=lfs -text
|
| 4 |
-
*.bz2 filter=lfs diff=lfs merge=lfs -text
|
| 5 |
-
*.ckpt filter=lfs diff=lfs merge=lfs -text
|
| 6 |
-
*.ftz filter=lfs diff=lfs merge=lfs -text
|
| 7 |
-
*.gz filter=lfs diff=lfs merge=lfs -text
|
| 8 |
-
*.h5 filter=lfs diff=lfs merge=lfs -text
|
| 9 |
-
*.joblib filter=lfs diff=lfs merge=lfs -text
|
| 10 |
-
*.lfs.* filter=lfs diff=lfs merge=lfs -text
|
| 11 |
-
*.mlmodel filter=lfs diff=lfs merge=lfs -text
|
| 12 |
-
*.model filter=lfs diff=lfs merge=lfs -text
|
| 13 |
-
*.msgpack filter=lfs diff=lfs merge=lfs -text
|
| 14 |
-
*.npy filter=lfs diff=lfs merge=lfs -text
|
| 15 |
-
*.npz filter=lfs diff=lfs merge=lfs -text
|
| 16 |
-
*.onnx filter=lfs diff=lfs merge=lfs -text
|
| 17 |
-
*.ot filter=lfs diff=lfs merge=lfs -text
|
| 18 |
-
*.parquet filter=lfs diff=lfs merge=lfs -text
|
| 19 |
-
*.pb filter=lfs diff=lfs merge=lfs -text
|
| 20 |
-
*.pickle filter=lfs diff=lfs merge=lfs -text
|
| 21 |
-
*.pkl filter=lfs diff=lfs merge=lfs -text
|
| 22 |
-
*.pt filter=lfs diff=lfs merge=lfs -text
|
| 23 |
-
*.pth filter=lfs diff=lfs merge=lfs -text
|
| 24 |
-
*.rar filter=lfs diff=lfs merge=lfs -text
|
| 25 |
*.safetensors filter=lfs diff=lfs merge=lfs -text
|
| 26 |
-
|
| 27 |
-
*.tar.* filter=lfs diff=lfs merge=lfs -text
|
| 28 |
-
*.tar filter=lfs diff=lfs merge=lfs -text
|
| 29 |
-
*.tflite filter=lfs diff=lfs merge=lfs -text
|
| 30 |
-
*.tgz filter=lfs diff=lfs merge=lfs -text
|
| 31 |
-
*.wasm filter=lfs diff=lfs merge=lfs -text
|
| 32 |
-
*.xz filter=lfs diff=lfs merge=lfs -text
|
| 33 |
-
*.zip filter=lfs diff=lfs merge=lfs -text
|
| 34 |
-
*.zst filter=lfs diff=lfs merge=lfs -text
|
| 35 |
-
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
*.safetensors filter=lfs diff=lfs merge=lfs -text
|
| 2 |
+
*.gguf filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
ESS-AIST-81M_q5_1.gguf
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1a527e514ccf313f064cb9014442945789ef816e437abf5b61d9f29c40eba448
|
| 3 |
+
size 121597856
|
ESS-AIST-81M_q8_0.gguf
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a9c79593801aa28823934aad09ec3efec5684b3802e294d004dbc71d93ca69b1
|
| 3 |
+
size 128732128
|
README.md
ADDED
|
@@ -0,0 +1,72 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
language:
|
| 3 |
+
- en
|
| 4 |
+
license: apache-2.0
|
| 5 |
+
tags:
|
| 6 |
+
- multimodal
|
| 7 |
+
- embedding
|
| 8 |
+
- trimodal
|
| 9 |
+
- retrieval
|
| 10 |
+
- image-text-audio
|
| 11 |
+
- feature-extraction
|
| 12 |
+
- gguf
|
| 13 |
+
- quantized
|
| 14 |
+
library_name: gguf
|
| 15 |
+
pipeline_tag: feature-extraction
|
| 16 |
+
datasets:
|
| 17 |
+
- custom
|
| 18 |
+
base_model:
|
| 19 |
+
- augmem/ESS-AIST-81M-preview
|
| 20 |
+
---
|
| 21 |
+
|
| 22 |
+
# ESS-AIST-81M Preview GGUF
|
| 23 |
+
|
| 24 |
+
This repository contains GGUF quantizations of `augmem/ESS-AIST-81M-preview`.
|
| 25 |
+
|
| 26 |
+
Base model:
|
| 27 |
+
- `augmem/ESS-AIST-81M-preview`
|
| 28 |
+
|
| 29 |
+
Quantizations:
|
| 30 |
+
- `ESS-AIST-81M_q8_0.gguf`
|
| 31 |
+
- `ESS-AIST-81M_q5_1.gguf`
|
| 32 |
+
|
| 33 |
+
These files were produced from the exact `checkpoint_epoch_11.pt` Cortext trial checkpoint.
|
| 34 |
+
|
| 35 |
+
## Exact Release Metrics
|
| 36 |
+
|
| 37 |
+
The quantized files correspond to the same release checkpoint and eval bundle as the base repo.
|
| 38 |
+
|
| 39 |
+
Speech holdout:
|
| 40 |
+
|
| 41 |
+
- `A->T_r1 = 0.4672`
|
| 42 |
+
- `T->A_r1 = 0.4606`
|
| 43 |
+
|
| 44 |
+
SALT:
|
| 45 |
+
|
| 46 |
+
- `I->T_r1 = 0.4149`
|
| 47 |
+
- `T->I_r1 = 0.4327`
|
| 48 |
+
- `A->T_r1 = 0.2408`
|
| 49 |
+
- `T->A_r1 = 0.2486`
|
| 50 |
+
|
| 51 |
+
Held-out ESS:
|
| 52 |
+
|
| 53 |
+
- `subject_key` same/different AUC: `0.5067`
|
| 54 |
+
- `event_key` same/different AUC: `0.8241`
|
| 55 |
+
|
| 56 |
+
## Files
|
| 57 |
+
|
| 58 |
+
| File | Purpose |
|
| 59 |
+
|---|---|
|
| 60 |
+
| `ESS-AIST-81M_q8_0.gguf` | Higher-accuracy GGUF |
|
| 61 |
+
| `ESS-AIST-81M_q5_1.gguf` | Smaller GGUF |
|
| 62 |
+
| `manifest.json` | Release manifest |
|
| 63 |
+
| `parameter_breakdown.json` | Exact parameter accounting |
|
| 64 |
+
| `retrieval_512_gt1030.json` | Exact 512d retrieval eval for the source checkpoint |
|
| 65 |
+
| `subject_eval.json` | Exact held-out subject eval for the source checkpoint |
|
| 66 |
+
| `event_eval.json` | Exact held-out event eval for the source checkpoint |
|
| 67 |
+
| `prefix_eval.json` | Prefix-level AUC summary |
|
| 68 |
+
|
| 69 |
+
## Notes
|
| 70 |
+
|
| 71 |
+
- This is a preview quantization repo for internal evaluation.
|
| 72 |
+
- The source checkpoint is still under active subject-data development.
|
event_eval.json
ADDED
|
@@ -0,0 +1,266 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"checkpoint": "/shared/augmem/triembed/checkpoints/ess_aist_full_v7_librispeech360_l4i/checkpoint_epoch_11.pt",
|
| 3 |
+
"split": "val",
|
| 4 |
+
"records_path": "/shared/augmem/triembed/checkpoints/ess_ait_86m_20260430T035907Z/ess_corpus_v7_subject_media_wit4096_speech100k_wavcaps100k_librispeech360/val.jsonl",
|
| 5 |
+
"views": {
|
| 6 |
+
"semantic_key": {
|
| 7 |
+
"event_same_different_auc": {
|
| 8 |
+
"auc": 0.829112461248993,
|
| 9 |
+
"positive_pairs": 7703,
|
| 10 |
+
"negative_pairs": 327075,
|
| 11 |
+
"positive_mean": 0.7533491437897564,
|
| 12 |
+
"negative_mean": 0.5995114385256625
|
| 13 |
+
},
|
| 14 |
+
"same_subject_different_event_rejection_auc": {
|
| 15 |
+
"auc": 0.5802306316888112,
|
| 16 |
+
"positive_pairs": 7703,
|
| 17 |
+
"negative_pairs": 118115,
|
| 18 |
+
"positive_mean": 0.7533491437897564,
|
| 19 |
+
"negative_mean": 0.7313196869598024
|
| 20 |
+
},
|
| 21 |
+
"stale_same_source_rejection_auc": {
|
| 22 |
+
"auc": null,
|
| 23 |
+
"positive_pairs": 7703,
|
| 24 |
+
"negative_pairs": 0,
|
| 25 |
+
"positive_mean": 0.7533491437897564,
|
| 26 |
+
"negative_mean": null
|
| 27 |
+
},
|
| 28 |
+
"wrong_active_rejection_auc": {
|
| 29 |
+
"auc": null,
|
| 30 |
+
"positive_pairs": 7703,
|
| 31 |
+
"negative_pairs": 0,
|
| 32 |
+
"positive_mean": 0.7533491437897564,
|
| 33 |
+
"negative_mean": null
|
| 34 |
+
},
|
| 35 |
+
"topic_shift_rejection_auc": {
|
| 36 |
+
"auc": 0.9697933441859231,
|
| 37 |
+
"positive_pairs": 7703,
|
| 38 |
+
"negative_pairs": 208960,
|
| 39 |
+
"positive_mean": 0.7533491437897564,
|
| 40 |
+
"negative_mean": 0.525006599016673
|
| 41 |
+
}
|
| 42 |
+
},
|
| 43 |
+
"subject_key": {
|
| 44 |
+
"event_same_different_auc": {
|
| 45 |
+
"auc": 0.6676734827239529,
|
| 46 |
+
"positive_pairs": 7703,
|
| 47 |
+
"negative_pairs": 327075,
|
| 48 |
+
"positive_mean": 0.668074605508422,
|
| 49 |
+
"negative_mean": 0.5629470272292642
|
| 50 |
+
},
|
| 51 |
+
"same_subject_different_event_rejection_auc": {
|
| 52 |
+
"auc": 0.1862661483021773,
|
| 53 |
+
"positive_pairs": 7703,
|
| 54 |
+
"negative_pairs": 118115,
|
| 55 |
+
"positive_mean": 0.668074605508422,
|
| 56 |
+
"negative_mean": 0.7863739344748515
|
| 57 |
+
},
|
| 58 |
+
"stale_same_source_rejection_auc": {
|
| 59 |
+
"auc": null,
|
| 60 |
+
"positive_pairs": 7703,
|
| 61 |
+
"negative_pairs": 0,
|
| 62 |
+
"positive_mean": 0.668074605508422,
|
| 63 |
+
"negative_mean": null
|
| 64 |
+
},
|
| 65 |
+
"wrong_active_rejection_auc": {
|
| 66 |
+
"auc": null,
|
| 67 |
+
"positive_pairs": 7703,
|
| 68 |
+
"negative_pairs": 0,
|
| 69 |
+
"positive_mean": 0.668074605508422,
|
| 70 |
+
"negative_mean": null
|
| 71 |
+
},
|
| 72 |
+
"topic_shift_rejection_auc": {
|
| 73 |
+
"auc": 0.9397898078829693,
|
| 74 |
+
"positive_pairs": 7703,
|
| 75 |
+
"negative_pairs": 208960,
|
| 76 |
+
"positive_mean": 0.668074605508422,
|
| 77 |
+
"negative_mean": 0.43665458298485105
|
| 78 |
+
}
|
| 79 |
+
},
|
| 80 |
+
"event_key": {
|
| 81 |
+
"event_same_different_auc": {
|
| 82 |
+
"auc": 0.8240710674869262,
|
| 83 |
+
"positive_pairs": 7703,
|
| 84 |
+
"negative_pairs": 327075,
|
| 85 |
+
"positive_mean": 0.6786398216704803,
|
| 86 |
+
"negative_mean": 0.4559661049066459
|
| 87 |
+
},
|
| 88 |
+
"same_subject_different_event_rejection_auc": {
|
| 89 |
+
"auc": 0.5534970574958717,
|
| 90 |
+
"positive_pairs": 7703,
|
| 91 |
+
"negative_pairs": 118115,
|
| 92 |
+
"positive_mean": 0.6786398216704803,
|
| 93 |
+
"negative_mean": 0.6430964619714105
|
| 94 |
+
},
|
| 95 |
+
"stale_same_source_rejection_auc": {
|
| 96 |
+
"auc": null,
|
| 97 |
+
"positive_pairs": 7703,
|
| 98 |
+
"negative_pairs": 0,
|
| 99 |
+
"positive_mean": 0.6786398216704803,
|
| 100 |
+
"negative_mean": null
|
| 101 |
+
},
|
| 102 |
+
"wrong_active_rejection_auc": {
|
| 103 |
+
"auc": null,
|
| 104 |
+
"positive_pairs": 7703,
|
| 105 |
+
"negative_pairs": 0,
|
| 106 |
+
"positive_mean": 0.6786398216704803,
|
| 107 |
+
"negative_mean": null
|
| 108 |
+
},
|
| 109 |
+
"topic_shift_rejection_auc": {
|
| 110 |
+
"auc": 0.9770134927840807,
|
| 111 |
+
"positive_pairs": 7703,
|
| 112 |
+
"negative_pairs": 208960,
|
| 113 |
+
"positive_mean": 0.6786398216704803,
|
| 114 |
+
"negative_mean": 0.35019034818428435
|
| 115 |
+
}
|
| 116 |
+
},
|
| 117 |
+
"full_key": {
|
| 118 |
+
"event_same_different_auc": {
|
| 119 |
+
"auc": 0.78280390304866,
|
| 120 |
+
"positive_pairs": 7703,
|
| 121 |
+
"negative_pairs": 327075,
|
| 122 |
+
"positive_mean": 0.7056915993491164,
|
| 123 |
+
"negative_mean": 0.5454881820918898
|
| 124 |
+
},
|
| 125 |
+
"same_subject_different_event_rejection_auc": {
|
| 126 |
+
"auc": 0.44637572176232837,
|
| 127 |
+
"positive_pairs": 7703,
|
| 128 |
+
"negative_pairs": 118115,
|
| 129 |
+
"positive_mean": 0.7056915993491164,
|
| 130 |
+
"negative_mean": 0.7219292155613277
|
| 131 |
+
},
|
| 132 |
+
"stale_same_source_rejection_auc": {
|
| 133 |
+
"auc": null,
|
| 134 |
+
"positive_pairs": 7703,
|
| 135 |
+
"negative_pairs": 0,
|
| 136 |
+
"positive_mean": 0.7056915993491164,
|
| 137 |
+
"negative_mean": null
|
| 138 |
+
},
|
| 139 |
+
"wrong_active_rejection_auc": {
|
| 140 |
+
"auc": null,
|
| 141 |
+
"positive_pairs": 7703,
|
| 142 |
+
"negative_pairs": 0,
|
| 143 |
+
"positive_mean": 0.7056915993491164,
|
| 144 |
+
"negative_mean": null
|
| 145 |
+
},
|
| 146 |
+
"topic_shift_rejection_auc": {
|
| 147 |
+
"auc": 0.9729705121252057,
|
| 148 |
+
"positive_pairs": 7703,
|
| 149 |
+
"negative_pairs": 208960,
|
| 150 |
+
"positive_mean": 0.7056915993491164,
|
| 151 |
+
"negative_mean": 0.4457545839475431
|
| 152 |
+
}
|
| 153 |
+
},
|
| 154 |
+
"prefix_512": {
|
| 155 |
+
"event_same_different_auc": {
|
| 156 |
+
"auc": 0.829112461248993,
|
| 157 |
+
"positive_pairs": 7703,
|
| 158 |
+
"negative_pairs": 327075,
|
| 159 |
+
"positive_mean": 0.7533491437897564,
|
| 160 |
+
"negative_mean": 0.5995114385256625
|
| 161 |
+
},
|
| 162 |
+
"same_subject_different_event_rejection_auc": {
|
| 163 |
+
"auc": 0.5802306316888112,
|
| 164 |
+
"positive_pairs": 7703,
|
| 165 |
+
"negative_pairs": 118115,
|
| 166 |
+
"positive_mean": 0.7533491437897564,
|
| 167 |
+
"negative_mean": 0.7313196869598024
|
| 168 |
+
},
|
| 169 |
+
"stale_same_source_rejection_auc": {
|
| 170 |
+
"auc": null,
|
| 171 |
+
"positive_pairs": 7703,
|
| 172 |
+
"negative_pairs": 0,
|
| 173 |
+
"positive_mean": 0.7533491437897564,
|
| 174 |
+
"negative_mean": null
|
| 175 |
+
},
|
| 176 |
+
"wrong_active_rejection_auc": {
|
| 177 |
+
"auc": null,
|
| 178 |
+
"positive_pairs": 7703,
|
| 179 |
+
"negative_pairs": 0,
|
| 180 |
+
"positive_mean": 0.7533491437897564,
|
| 181 |
+
"negative_mean": null
|
| 182 |
+
},
|
| 183 |
+
"topic_shift_rejection_auc": {
|
| 184 |
+
"auc": 0.9697933441859231,
|
| 185 |
+
"positive_pairs": 7703,
|
| 186 |
+
"negative_pairs": 208960,
|
| 187 |
+
"positive_mean": 0.7533491437897564,
|
| 188 |
+
"negative_mean": 0.525006599016673
|
| 189 |
+
}
|
| 190 |
+
},
|
| 191 |
+
"prefix_1024": {
|
| 192 |
+
"event_same_different_auc": {
|
| 193 |
+
"auc": 0.7453008223026156,
|
| 194 |
+
"positive_pairs": 7703,
|
| 195 |
+
"negative_pairs": 327075,
|
| 196 |
+
"positive_mean": 0.7183707235626442,
|
| 197 |
+
"negative_mean": 0.5870387001189491
|
| 198 |
+
},
|
| 199 |
+
"same_subject_different_event_rejection_auc": {
|
| 200 |
+
"auc": 0.3570483627258597,
|
| 201 |
+
"positive_pairs": 7703,
|
| 202 |
+
"negative_pairs": 118115,
|
| 203 |
+
"positive_mean": 0.7183707235626442,
|
| 204 |
+
"negative_mean": 0.7610074661872391
|
| 205 |
+
},
|
| 206 |
+
"stale_same_source_rejection_auc": {
|
| 207 |
+
"auc": null,
|
| 208 |
+
"positive_pairs": 7703,
|
| 209 |
+
"negative_pairs": 0,
|
| 210 |
+
"positive_mean": 0.7183707235626442,
|
| 211 |
+
"negative_mean": null
|
| 212 |
+
},
|
| 213 |
+
"wrong_active_rejection_auc": {
|
| 214 |
+
"auc": null,
|
| 215 |
+
"positive_pairs": 7703,
|
| 216 |
+
"negative_pairs": 0,
|
| 217 |
+
"positive_mean": 0.7183707235626442,
|
| 218 |
+
"negative_mean": null
|
| 219 |
+
},
|
| 220 |
+
"topic_shift_rejection_auc": {
|
| 221 |
+
"auc": 0.9647611939666115,
|
| 222 |
+
"positive_pairs": 7703,
|
| 223 |
+
"negative_pairs": 208960,
|
| 224 |
+
"positive_mean": 0.7183707235626442,
|
| 225 |
+
"negative_mean": 0.48870255538236756
|
| 226 |
+
}
|
| 227 |
+
},
|
| 228 |
+
"prefix_1536": {
|
| 229 |
+
"event_same_different_auc": {
|
| 230 |
+
"auc": 0.78280390304866,
|
| 231 |
+
"positive_pairs": 7703,
|
| 232 |
+
"negative_pairs": 327075,
|
| 233 |
+
"positive_mean": 0.7056915993491164,
|
| 234 |
+
"negative_mean": 0.5454881820918898
|
| 235 |
+
},
|
| 236 |
+
"same_subject_different_event_rejection_auc": {
|
| 237 |
+
"auc": 0.44637572176232837,
|
| 238 |
+
"positive_pairs": 7703,
|
| 239 |
+
"negative_pairs": 118115,
|
| 240 |
+
"positive_mean": 0.7056915993491164,
|
| 241 |
+
"negative_mean": 0.7219292155613277
|
| 242 |
+
},
|
| 243 |
+
"stale_same_source_rejection_auc": {
|
| 244 |
+
"auc": null,
|
| 245 |
+
"positive_pairs": 7703,
|
| 246 |
+
"negative_pairs": 0,
|
| 247 |
+
"positive_mean": 0.7056915993491164,
|
| 248 |
+
"negative_mean": null
|
| 249 |
+
},
|
| 250 |
+
"wrong_active_rejection_auc": {
|
| 251 |
+
"auc": null,
|
| 252 |
+
"positive_pairs": 7703,
|
| 253 |
+
"negative_pairs": 0,
|
| 254 |
+
"positive_mean": 0.7056915993491164,
|
| 255 |
+
"negative_mean": null
|
| 256 |
+
},
|
| 257 |
+
"topic_shift_rejection_auc": {
|
| 258 |
+
"auc": 0.9729705121252057,
|
| 259 |
+
"positive_pairs": 7703,
|
| 260 |
+
"negative_pairs": 208960,
|
| 261 |
+
"positive_mean": 0.7056915993491164,
|
| 262 |
+
"negative_mean": 0.4457545839475431
|
| 263 |
+
}
|
| 264 |
+
}
|
| 265 |
+
}
|
| 266 |
+
}
|
export_metadata.json
ADDED
|
@@ -0,0 +1,49 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model_name": "ESS-AIST-81M",
|
| 3 |
+
"base_family": "AIST-86M-compatible",
|
| 4 |
+
"output_dimension": 1536,
|
| 5 |
+
"slices": {
|
| 6 |
+
"semantic": [
|
| 7 |
+
0,
|
| 8 |
+
512
|
| 9 |
+
],
|
| 10 |
+
"subject": [
|
| 11 |
+
512,
|
| 12 |
+
1024
|
| 13 |
+
],
|
| 14 |
+
"event": [
|
| 15 |
+
1024,
|
| 16 |
+
1536
|
| 17 |
+
]
|
| 18 |
+
},
|
| 19 |
+
"prefixes": {
|
| 20 |
+
"semantic_prefix": 512,
|
| 21 |
+
"semantic_subject_prefix": 1024,
|
| 22 |
+
"full_prefix": 1536
|
| 23 |
+
},
|
| 24 |
+
"normalized_views": {
|
| 25 |
+
"semantic_key": "l2norm(z[0:512])",
|
| 26 |
+
"subject_key": "l2norm(z[512:1024])",
|
| 27 |
+
"event_key": "l2norm(z[1024:1536])",
|
| 28 |
+
"full_key": "l2norm(z[0:1536])"
|
| 29 |
+
},
|
| 30 |
+
"supported_modalities": [
|
| 31 |
+
"text",
|
| 32 |
+
"image",
|
| 33 |
+
"audio"
|
| 34 |
+
],
|
| 35 |
+
"normalization_behavior": {
|
| 36 |
+
"raw_embedding": "un-normalized 1536d vector",
|
| 37 |
+
"recommended_runtime": "L2-normalize per-slice or full vector depending on task"
|
| 38 |
+
},
|
| 39 |
+
"matryoshka_behavior": {
|
| 40 |
+
"512": "semantic retrieval",
|
| 41 |
+
"1024": "semantic plus subject continuity",
|
| 42 |
+
"1536": "semantic plus subject plus event continuity"
|
| 43 |
+
},
|
| 44 |
+
"optional_probes": [
|
| 45 |
+
"salience_score",
|
| 46 |
+
"novelty_score",
|
| 47 |
+
"boundary_score"
|
| 48 |
+
]
|
| 49 |
+
}
|
manifest.json
ADDED
|
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model_id": "ESS-AIST-81M",
|
| 3 |
+
"trimodal_checkpoint": "/shared/augmem/triembed/checkpoints/ess_aist_full_v7_librispeech360_l4i/checkpoint_epoch_11.pt",
|
| 4 |
+
"audio_checkpoint": "/shared/augmem/triembed/checkpoints/mn20_native_lora_aistmix_audioheavy100k175k175k_continue_from_balanced_20260426T143137Z/latest_model.pt",
|
| 5 |
+
"safetensors": "/shared/augmem/triembed/dist/ESS-AIST-81M-preview/ESS-AIST-81M.safetensors",
|
| 6 |
+
"gguf": [
|
| 7 |
+
"/shared/augmem/triembed/dist/ESS-AIST-81M-preview/ESS-AIST-81M_q8_0.gguf",
|
| 8 |
+
"/shared/augmem/triembed/dist/ESS-AIST-81M-preview/ESS-AIST-81M_q5_1.gguf"
|
| 9 |
+
]
|
| 10 |
+
}
|
parameter_breakdown.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"text_encoder": 22861056,
|
| 3 |
+
"image_encoder": 8434512,
|
| 4 |
+
"audio_encoder": 20639974,
|
| 5 |
+
"image_projection": 9975296,
|
| 6 |
+
"audio_projection": 9975296,
|
| 7 |
+
"text_projection": 8926720,
|
| 8 |
+
"total_exact_loaded_params": 80812854
|
| 9 |
+
}
|
prefix_eval.json
ADDED
|
@@ -0,0 +1,48 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"checkpoint": "/shared/augmem/triembed/checkpoints/ess_aist_full_v7_librispeech360_l4i/checkpoint_epoch_11.pt",
|
| 3 |
+
"split": "val",
|
| 4 |
+
"views": {
|
| 5 |
+
"semantic_key": {
|
| 6 |
+
"subject_same_different_auc": 0.4265240470767738,
|
| 7 |
+
"event_same_different_auc": 0.829112461248993,
|
| 8 |
+
"same_topic_different_subject_rejection_auc": 0.4265240470767738,
|
| 9 |
+
"same_subject_different_event_rejection_auc": 0.5802306316888112
|
| 10 |
+
},
|
| 11 |
+
"subject_key": {
|
| 12 |
+
"subject_same_different_auc": 0.5066875746523821,
|
| 13 |
+
"event_same_different_auc": 0.6676734827239529,
|
| 14 |
+
"same_topic_different_subject_rejection_auc": 0.5066875746523821,
|
| 15 |
+
"same_subject_different_event_rejection_auc": 0.1862661483021773
|
| 16 |
+
},
|
| 17 |
+
"event_key": {
|
| 18 |
+
"subject_same_different_auc": 0.3832485276953712,
|
| 19 |
+
"event_same_different_auc": 0.8240710674869262,
|
| 20 |
+
"same_topic_different_subject_rejection_auc": 0.3832485276953712,
|
| 21 |
+
"same_subject_different_event_rejection_auc": 0.5534970574958717
|
| 22 |
+
},
|
| 23 |
+
"full_key": {
|
| 24 |
+
"subject_same_different_auc": 0.42067046032727157,
|
| 25 |
+
"event_same_different_auc": 0.78280390304866,
|
| 26 |
+
"same_topic_different_subject_rejection_auc": 0.42067046032727157,
|
| 27 |
+
"same_subject_different_event_rejection_auc": 0.44637572176232837
|
| 28 |
+
},
|
| 29 |
+
"prefix_512": {
|
| 30 |
+
"subject_same_different_auc": 0.4265240470767738,
|
| 31 |
+
"event_same_different_auc": 0.829112461248993,
|
| 32 |
+
"same_topic_different_subject_rejection_auc": 0.4265240470767738,
|
| 33 |
+
"same_subject_different_event_rejection_auc": 0.5802306316888112
|
| 34 |
+
},
|
| 35 |
+
"prefix_1024": {
|
| 36 |
+
"subject_same_different_auc": 0.4690923681629257,
|
| 37 |
+
"event_same_different_auc": 0.7453008223026156,
|
| 38 |
+
"same_topic_different_subject_rejection_auc": 0.4690923681629257,
|
| 39 |
+
"same_subject_different_event_rejection_auc": 0.3570483627258597
|
| 40 |
+
},
|
| 41 |
+
"prefix_1536": {
|
| 42 |
+
"subject_same_different_auc": 0.42067046032727157,
|
| 43 |
+
"event_same_different_auc": 0.78280390304866,
|
| 44 |
+
"same_topic_different_subject_rejection_auc": 0.42067046032727157,
|
| 45 |
+
"same_subject_different_event_rejection_auc": 0.44637572176232837
|
| 46 |
+
}
|
| 47 |
+
}
|
| 48 |
+
}
|
retrieval_512_gt1030.json
ADDED
|
@@ -0,0 +1,40 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"SALT-512": {
|
| 3 |
+
"A->I_r1": 0.4828965961933136,
|
| 4 |
+
"A->I_r10": 0.8761752843856812,
|
| 5 |
+
"A->I_r5": 0.7863572835922241,
|
| 6 |
+
"A->T_r1": 0.24084816873073578,
|
| 7 |
+
"A->T_r10": 0.5153030753135681,
|
| 8 |
+
"A->T_r5": 0.45209044218063354,
|
| 9 |
+
"I->A_r1": 0.46209242939949036,
|
| 10 |
+
"I->A_r10": 0.881176233291626,
|
| 11 |
+
"I->A_r5": 0.7905581593513489,
|
| 12 |
+
"I->T_r1": 0.41488298773765564,
|
| 13 |
+
"I->T_r10": 0.5707141757011414,
|
| 14 |
+
"I->T_r5": 0.5401080250740051,
|
| 15 |
+
"T->A_r1": 0.2486497312784195,
|
| 16 |
+
"T->A_r10": 0.5323064923286438,
|
| 17 |
+
"T->A_r5": 0.46209242939949036,
|
| 18 |
+
"T->I_r1": 0.43268653750419617,
|
| 19 |
+
"T->I_r10": 0.5763152837753296,
|
| 20 |
+
"T->I_r5": 0.550710141658783
|
| 21 |
+
},
|
| 22 |
+
"_meta": {
|
| 23 |
+
"audio_suffix": "mn20_audioheavy_lora1280_audio_features",
|
| 24 |
+
"checkpoint": "/shared/augmem/triembed/checkpoints/ess_aist_full_v7_librispeech360_l4i/checkpoint_epoch_11.pt",
|
| 25 |
+
"device": "NVIDIA GeForce GT 1030",
|
| 26 |
+
"dims": [
|
| 27 |
+
512
|
| 28 |
+
],
|
| 29 |
+
"encoder_name": "mobilenetv4_conv_medium",
|
| 30 |
+
"image_suffix": "mobilenetv4_conv_medium_image_features"
|
| 31 |
+
},
|
| 32 |
+
"speech_chatterbox-512": {
|
| 33 |
+
"A->T_r1": 0.46719998121261597,
|
| 34 |
+
"A->T_r10": 0.824999988079071,
|
| 35 |
+
"A->T_r5": 0.739799976348877,
|
| 36 |
+
"T->A_r1": 0.46059998869895935,
|
| 37 |
+
"T->A_r10": 0.8277999758720398,
|
| 38 |
+
"T->A_r5": 0.7425999641418457
|
| 39 |
+
}
|
| 40 |
+
}
|
subject_eval.json
ADDED
|
@@ -0,0 +1,119 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"checkpoint": "/shared/augmem/triembed/checkpoints/ess_aist_full_v7_librispeech360_l4i/checkpoint_epoch_11.pt",
|
| 3 |
+
"split": "val",
|
| 4 |
+
"records_path": "/shared/augmem/triembed/checkpoints/ess_ait_86m_20260430T035907Z/ess_corpus_v7_subject_media_wit4096_speech100k_wavcaps100k_librispeech360/val.jsonl",
|
| 5 |
+
"views": {
|
| 6 |
+
"semantic_key": {
|
| 7 |
+
"subject_same_different_auc": {
|
| 8 |
+
"auc": 0.4265240470767738,
|
| 9 |
+
"positive_pairs": 160248,
|
| 10 |
+
"negative_pairs": 6805,
|
| 11 |
+
"positive_mean": 0.7436165443684017,
|
| 12 |
+
"negative_mean": 0.7611866422867968
|
| 13 |
+
},
|
| 14 |
+
"same_topic_different_subject_rejection_auc": {
|
| 15 |
+
"auc": 0.4265240470767738,
|
| 16 |
+
"positive_pairs": 160248,
|
| 17 |
+
"negative_pairs": 6805,
|
| 18 |
+
"positive_mean": 0.7436165443684017,
|
| 19 |
+
"negative_mean": 0.7611866422867968
|
| 20 |
+
}
|
| 21 |
+
},
|
| 22 |
+
"subject_key": {
|
| 23 |
+
"subject_same_different_auc": {
|
| 24 |
+
"auc": 0.5066875746523821,
|
| 25 |
+
"positive_pairs": 160248,
|
| 26 |
+
"negative_pairs": 6805,
|
| 27 |
+
"positive_mean": 0.7964573271532047,
|
| 28 |
+
"negative_mean": 0.7948588548339001
|
| 29 |
+
},
|
| 30 |
+
"same_topic_different_subject_rejection_auc": {
|
| 31 |
+
"auc": 0.5066875746523821,
|
| 32 |
+
"positive_pairs": 160248,
|
| 33 |
+
"negative_pairs": 6805,
|
| 34 |
+
"positive_mean": 0.7964573271532047,
|
| 35 |
+
"negative_mean": 0.7948588548339001
|
| 36 |
+
}
|
| 37 |
+
},
|
| 38 |
+
"event_key": {
|
| 39 |
+
"subject_same_different_auc": {
|
| 40 |
+
"auc": 0.3832485276953712,
|
| 41 |
+
"positive_pairs": 160248,
|
| 42 |
+
"negative_pairs": 6805,
|
| 43 |
+
"positive_mean": 0.6533856675037972,
|
| 44 |
+
"negative_mean": 0.7097943563909324
|
| 45 |
+
},
|
| 46 |
+
"same_topic_different_subject_rejection_auc": {
|
| 47 |
+
"auc": 0.3832485276953712,
|
| 48 |
+
"positive_pairs": 160248,
|
| 49 |
+
"negative_pairs": 6805,
|
| 50 |
+
"positive_mean": 0.6533856675037972,
|
| 51 |
+
"negative_mean": 0.7097943563909324
|
| 52 |
+
}
|
| 53 |
+
},
|
| 54 |
+
"full_key": {
|
| 55 |
+
"subject_same_different_auc": {
|
| 56 |
+
"auc": 0.42067046032727157,
|
| 57 |
+
"positive_pairs": 160248,
|
| 58 |
+
"negative_pairs": 6805,
|
| 59 |
+
"positive_mean": 0.7333961783599913,
|
| 60 |
+
"negative_mean": 0.754728921120172
|
| 61 |
+
},
|
| 62 |
+
"same_topic_different_subject_rejection_auc": {
|
| 63 |
+
"auc": 0.42067046032727157,
|
| 64 |
+
"positive_pairs": 160248,
|
| 65 |
+
"negative_pairs": 6805,
|
| 66 |
+
"positive_mean": 0.7333961783599913,
|
| 67 |
+
"negative_mean": 0.754728921120172
|
| 68 |
+
}
|
| 69 |
+
},
|
| 70 |
+
"prefix_512": {
|
| 71 |
+
"subject_same_different_auc": {
|
| 72 |
+
"auc": 0.4265240470767738,
|
| 73 |
+
"positive_pairs": 160248,
|
| 74 |
+
"negative_pairs": 6805,
|
| 75 |
+
"positive_mean": 0.7436165443684017,
|
| 76 |
+
"negative_mean": 0.7611866422867968
|
| 77 |
+
},
|
| 78 |
+
"same_topic_different_subject_rejection_auc": {
|
| 79 |
+
"auc": 0.4265240470767738,
|
| 80 |
+
"positive_pairs": 160248,
|
| 81 |
+
"negative_pairs": 6805,
|
| 82 |
+
"positive_mean": 0.7436165443684017,
|
| 83 |
+
"negative_mean": 0.7611866422867968
|
| 84 |
+
}
|
| 85 |
+
},
|
| 86 |
+
"prefix_1024": {
|
| 87 |
+
"subject_same_different_auc": {
|
| 88 |
+
"auc": 0.4690923681629257,
|
| 89 |
+
"positive_pairs": 160248,
|
| 90 |
+
"negative_pairs": 6805,
|
| 91 |
+
"positive_mean": 0.7721689081039962,
|
| 92 |
+
"negative_mean": 0.7791595536982812
|
| 93 |
+
},
|
| 94 |
+
"same_topic_different_subject_rejection_auc": {
|
| 95 |
+
"auc": 0.4690923681629257,
|
| 96 |
+
"positive_pairs": 160248,
|
| 97 |
+
"negative_pairs": 6805,
|
| 98 |
+
"positive_mean": 0.7721689081039962,
|
| 99 |
+
"negative_mean": 0.7791595536982812
|
| 100 |
+
}
|
| 101 |
+
},
|
| 102 |
+
"prefix_1536": {
|
| 103 |
+
"subject_same_different_auc": {
|
| 104 |
+
"auc": 0.42067046032727157,
|
| 105 |
+
"positive_pairs": 160248,
|
| 106 |
+
"negative_pairs": 6805,
|
| 107 |
+
"positive_mean": 0.7333961783599913,
|
| 108 |
+
"negative_mean": 0.754728921120172
|
| 109 |
+
},
|
| 110 |
+
"same_topic_different_subject_rejection_auc": {
|
| 111 |
+
"auc": 0.42067046032727157,
|
| 112 |
+
"positive_pairs": 160248,
|
| 113 |
+
"negative_pairs": 6805,
|
| 114 |
+
"positive_mean": 0.7333961783599913,
|
| 115 |
+
"negative_mean": 0.754728921120172
|
| 116 |
+
}
|
| 117 |
+
}
|
| 118 |
+
}
|
| 119 |
+
}
|