Token Classification
Transformers
Safetensors
modernbert
semantic-router
vela
hallucination-detection
Instructions to use vllm-sr/Vela-1.0-Encoder-307M-Halu with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use vllm-sr/Vela-1.0-Encoder-307M-Halu with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("token-classification", model="vllm-sr/Vela-1.0-Encoder-307M-Halu")# pip install -U transformers accelerate # Load model directly from transformers import AutoTokenizer, AutoModelForTokenClassification tokenizer = AutoTokenizer.from_pretrained("vllm-sr/Vela-1.0-Encoder-307M-Halu") model = AutoModelForTokenClassification.from_pretrained("vllm-sr/Vela-1.0-Encoder-307M-Halu", device_map="auto") - Notebooks
- Google Colab
- Kaggle
Publish Vela Halu model and evaluation scores
Browse files- README.md +7 -5
- model.safetensors +1 -1
- scores.json +64 -64
README.md
CHANGED
|
@@ -32,16 +32,18 @@ Vela Halu finds answer spans unsupported by the supplied evidence, for grounded
|
|
| 32 |
|
| 33 |
Labels are `supported` (0) and `hallucinated` (1). Spans use Unicode character offsets in the answer. Evidence support is distinct from real-world factual truth; an empty result does not guarantee correctness.
|
| 34 |
|
|
|
|
|
|
|
| 35 |
## Evaluation
|
| 36 |
|
| 37 |
-
Scores (0-100) on the 10,698-example
|
| 38 |
|
| 39 |
| Metric | Vela Halu |
|
| 40 |
|---|---:|
|
| 41 |
-
| Overall span F1 |
|
| 42 |
-
| Overall example F1 | 87.
|
| 43 |
-
| Code span F1 |
|
| 44 |
-
| Tool output span F1 |
|
| 45 |
|
| 46 |
Span F1 measures character overlap; example F1 measures whether an answer contains any hallucinated span. Evaluation uses bfloat16, an 8,192-token limit, `only_first` pair truncation, and token scores strictly above 0.5. [Full scores](./scores.json) include all source groups and truncation statistics.
|
| 47 |
|
|
|
|
| 32 |
|
| 33 |
Labels are `supported` (0) and `hallucinated` (1). Spans use Unicode character offsets in the answer. Evidence support is distinct from real-world factual truth; an empty result does not guarantee correctness.
|
| 34 |
|
| 35 |
+
Answers requiring arithmetic or multi-step reasoning beyond explicit context may be incorrectly flagged.
|
| 36 |
+
|
| 37 |
## Evaluation
|
| 38 |
|
| 39 |
+
Scores (0-100) on the 10,698-example fixed evaluation set. Higher is better.
|
| 40 |
|
| 41 |
| Metric | Vela Halu |
|
| 42 |
|---|---:|
|
| 43 |
+
| Overall span F1 | 64.68 |
|
| 44 |
+
| Overall example F1 | 87.49 |
|
| 45 |
+
| Code span F1 | 53.04 |
|
| 46 |
+
| Tool output span F1 | 64.28 |
|
| 47 |
|
| 48 |
Span F1 measures character overlap; example F1 measures whether an answer contains any hallucinated span. Evaluation uses bfloat16, an 8,192-token limit, `only_first` pair truncation, and token scores strictly above 0.5. [Full scores](./scores.json) include all source groups and truncation statistics.
|
| 49 |
|
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1230141424
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0f51bf4a6e1462e88733c36e1e59fa96c1188c8c470de7829b51c95298247be6
|
| 3 |
size 1230141424
|
scores.json
CHANGED
|
@@ -9,92 +9,92 @@
|
|
| 9 |
"threshold": 0.5,
|
| 10 |
"overall": {
|
| 11 |
"n": 10698,
|
| 12 |
-
"span_f1": 0.
|
| 13 |
-
"span_p": 0.
|
| 14 |
-
"span_r": 0.
|
| 15 |
-
"ex_f1": 0.
|
| 16 |
-
"ex_p": 0.
|
| 17 |
-
"ex_r": 0.
|
| 18 |
-
"iou": 0.
|
| 19 |
-
"clean_fpr": 0.
|
| 20 |
},
|
| 21 |
"by_source": {
|
| 22 |
"lettucedetect-acl": {
|
| 23 |
"n": 440,
|
| 24 |
-
"span_f1": 0.
|
| 25 |
-
"span_p": 0.
|
| 26 |
-
"span_r": 0.
|
| 27 |
-
"ex_f1": 0.
|
| 28 |
-
"ex_p": 0.
|
| 29 |
-
"ex_r": 0.
|
| 30 |
-
"iou": 0.
|
| 31 |
-
"clean_fpr": 0.
|
| 32 |
},
|
| 33 |
"lettucedetect-code-agent": {
|
| 34 |
"n": 2015,
|
| 35 |
-
"span_f1": 0.
|
| 36 |
-
"span_p": 0.
|
| 37 |
-
"span_r": 0.
|
| 38 |
-
"ex_f1": 0.
|
| 39 |
-
"ex_p": 0.
|
| 40 |
-
"ex_r": 0.
|
| 41 |
-
"iou": 0.
|
| 42 |
-
"clean_fpr": 0.
|
| 43 |
},
|
| 44 |
"lettucedetect-readme": {
|
| 45 |
"n": 641,
|
| 46 |
-
"span_f1": 0.
|
| 47 |
-
"span_p": 0.
|
| 48 |
-
"span_r": 0.
|
| 49 |
-
"ex_f1": 0.
|
| 50 |
-
"ex_p": 0.
|
| 51 |
-
"ex_r": 0.
|
| 52 |
-
"iou": 0.
|
| 53 |
-
"clean_fpr": 0.
|
| 54 |
},
|
| 55 |
"lettucedetect-tool-output": {
|
| 56 |
"n": 617,
|
| 57 |
-
"span_f1": 0.
|
| 58 |
-
"span_p": 0.
|
| 59 |
-
"span_r": 0.
|
| 60 |
-
"ex_f1": 0.
|
| 61 |
-
"ex_p": 0.
|
| 62 |
-
"ex_r": 0.
|
| 63 |
-
"iou": 0.
|
| 64 |
-
"clean_fpr": 0.
|
| 65 |
},
|
| 66 |
"lettucedetect-wikipedia": {
|
| 67 |
"n": 1388,
|
| 68 |
-
"span_f1": 0.
|
| 69 |
-
"span_p": 0.
|
| 70 |
-
"span_r": 0.
|
| 71 |
-
"ex_f1": 0.
|
| 72 |
-
"ex_p": 0.
|
| 73 |
-
"ex_r": 0.
|
| 74 |
-
"iou": 0.
|
| 75 |
-
"clean_fpr": 0.
|
| 76 |
},
|
| 77 |
"psiloqa": {
|
| 78 |
"n": 2897,
|
| 79 |
-
"span_f1": 0.
|
| 80 |
-
"span_p": 0.
|
| 81 |
-
"span_r": 0.
|
| 82 |
-
"ex_f1": 0.
|
| 83 |
-
"ex_p": 0.
|
| 84 |
-
"ex_r": 0.
|
| 85 |
-
"iou": 0.
|
| 86 |
-
"clean_fpr": 0.
|
| 87 |
},
|
| 88 |
"ragtruth": {
|
| 89 |
"n": 2700,
|
| 90 |
-
"span_f1": 0.
|
| 91 |
-
"span_p": 0.
|
| 92 |
-
"span_r": 0.
|
| 93 |
-
"ex_f1": 0.
|
| 94 |
-
"ex_p": 0.
|
| 95 |
-
"ex_r": 0.
|
| 96 |
-
"iou": 0.
|
| 97 |
-
"clean_fpr": 0.
|
| 98 |
}
|
| 99 |
},
|
| 100 |
"truncation": {
|
|
|
|
| 9 |
"threshold": 0.5,
|
| 10 |
"overall": {
|
| 11 |
"n": 10698,
|
| 12 |
+
"span_f1": 0.6468063721645629,
|
| 13 |
+
"span_p": 0.7001953973777478,
|
| 14 |
+
"span_r": 0.6009822273490223,
|
| 15 |
+
"ex_f1": 0.8749271864858118,
|
| 16 |
+
"ex_p": 0.8913190912173619,
|
| 17 |
+
"ex_r": 0.8591273083837229,
|
| 18 |
+
"iou": 0.6791407403861177,
|
| 19 |
+
"clean_fpr": 0.13998689670233674
|
| 20 |
},
|
| 21 |
"by_source": {
|
| 22 |
"lettucedetect-acl": {
|
| 23 |
"n": 440,
|
| 24 |
+
"span_f1": 0.6033354880413704,
|
| 25 |
+
"span_p": 0.720407533189256,
|
| 26 |
+
"span_r": 0.5189946619217082,
|
| 27 |
+
"ex_f1": 0.8590021691973969,
|
| 28 |
+
"ex_p": 0.8215767634854771,
|
| 29 |
+
"ex_r": 0.9,
|
| 30 |
+
"iou": 0.6506637226712354,
|
| 31 |
+
"clean_fpr": 0.19545454545454546
|
| 32 |
},
|
| 33 |
"lettucedetect-code-agent": {
|
| 34 |
"n": 2015,
|
| 35 |
+
"span_f1": 0.5304085445334227,
|
| 36 |
+
"span_p": 0.6548436524601328,
|
| 37 |
+
"span_r": 0.44571299290373134,
|
| 38 |
+
"ex_f1": 0.7826510721247564,
|
| 39 |
+
"ex_p": 0.7736030828516378,
|
| 40 |
+
"ex_r": 0.7919132149901381,
|
| 41 |
+
"iou": 0.5799284591741568,
|
| 42 |
+
"clean_fpr": 0.23476523476523475
|
| 43 |
},
|
| 44 |
"lettucedetect-readme": {
|
| 45 |
"n": 641,
|
| 46 |
+
"span_f1": 0.7668571829306778,
|
| 47 |
+
"span_p": 0.8180594000149622,
|
| 48 |
+
"span_r": 0.7216869060190074,
|
| 49 |
+
"ex_f1": 0.9153605015673981,
|
| 50 |
+
"ex_p": 0.9182389937106918,
|
| 51 |
+
"ex_r": 0.9125,
|
| 52 |
+
"iou": 0.7844094959106822,
|
| 53 |
+
"clean_fpr": 0.08099688473520249
|
| 54 |
},
|
| 55 |
"lettucedetect-tool-output": {
|
| 56 |
"n": 617,
|
| 57 |
+
"span_f1": 0.6427959358550618,
|
| 58 |
+
"span_p": 0.7796585003711952,
|
| 59 |
+
"span_r": 0.5468082890763303,
|
| 60 |
+
"ex_f1": 0.8273504273504273,
|
| 61 |
+
"ex_p": 0.8736462093862816,
|
| 62 |
+
"ex_r": 0.7857142857142857,
|
| 63 |
+
"iou": 0.7027343010780659,
|
| 64 |
+
"clean_fpr": 0.11326860841423948
|
| 65 |
},
|
| 66 |
"lettucedetect-wikipedia": {
|
| 67 |
"n": 1388,
|
| 68 |
+
"span_f1": 0.7343291203881016,
|
| 69 |
+
"span_p": 0.7708389171803806,
|
| 70 |
+
"span_r": 0.7011214041930766,
|
| 71 |
+
"ex_f1": 0.9298369950389794,
|
| 72 |
+
"ex_p": 0.9149232914923291,
|
| 73 |
+
"ex_r": 0.9452449567723343,
|
| 74 |
+
"iou": 0.7905999840786986,
|
| 75 |
+
"clean_fpr": 0.08789625360230548
|
| 76 |
},
|
| 77 |
"psiloqa": {
|
| 78 |
"n": 2897,
|
| 79 |
+
"span_f1": 0.7084256755107189,
|
| 80 |
+
"span_p": 0.7054881112533835,
|
| 81 |
+
"span_r": 0.7113878053564034,
|
| 82 |
+
"ex_f1": 0.9406055900621118,
|
| 83 |
+
"ex_p": 0.9569510268562401,
|
| 84 |
+
"ex_r": 0.9248091603053435,
|
| 85 |
+
"iou": 0.6254885367916755,
|
| 86 |
+
"clean_fpr": 0.3935018050541516
|
| 87 |
},
|
| 88 |
"ragtruth": {
|
| 89 |
"n": 2700,
|
| 90 |
+
"span_f1": 0.5254962135743007,
|
| 91 |
+
"span_p": 0.6738326440634735,
|
| 92 |
+
"span_r": 0.43068574822129324,
|
| 93 |
+
"ex_f1": 0.7485448195576252,
|
| 94 |
+
"ex_p": 0.8296774193548387,
|
| 95 |
+
"ex_r": 0.6818663838812301,
|
| 96 |
+
"iou": 0.727708569558819,
|
| 97 |
+
"clean_fpr": 0.07512805919180421
|
| 98 |
}
|
| 99 |
},
|
| 100 |
"truncation": {
|