INDEX

Explanations

phrases indicating a request for trustworthy information or feedback

oai_token-act-pair · gpt-3.5-turbo

questions regarding trustworthiness and subscriptions to news content

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

beit

-0.77

naire

-0.67

 misunder

-0.63

alist

-0.61

esome

-0.60

hof

-0.60

liest

-0.59

eni

-0.58

 Calais

-0.57

 helicop

-0.57

POSITIVE LOGITS

 Email

0.91

utm

0.90

 Subscribe

0.83

 Attend

0.78

 Become

0.74

 Replay

0.74

Content

0.71

 Want

0.71

 Visit

0.70

Try

0.69

Activations Density 0.021%

phrases indicating a request for trustworthy information or feedback

questions regarding trustworthiness and subscriptions to news content

No Comments

No Known Activations