INDEX

Explanations

instances where permission is mentioned or required

oai_token-act-pair · gpt-3.5-turbo

references to authorization and consent

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.0.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

imated

-0.61

rose

-0.61

arde

-0.61

 contrasted

-0.60

 Trend

-0.60

Conclusion

-0.59

igue

-0.59

 summed

-0.58

 compared

-0.57

iants

-0.57

POSITIVE LOGITS

 permission

3.93

 permissions

2.22

 consent

2.02

 authorization

1.92

 approval

1.78

 Consent

1.58

 permits

1.44

 blessing

1.37

 licence

1.33

 permit

1.30

Activations Density 0.015%

instances where permission is mentioned or required

references to authorization and consent

No Comments

No Known Activations