INDEX

Explanations

phrases related to investigations and probes conducted by authorities

oai_token-act-pair · gpt-3.5-turbo

phrases related to investigations and oversight processes

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 toast

-0.77

 ç¥ŀ

-0.75

 oranges

-0.74

uously

-0.72

ibo

-0.69

 glide

-0.68

utra

-0.67

tall

-0.65

veland

-0.63

izens

-0.62

POSITIVE LOGITS

 prosecut

0.85

 investigative

0.80

 prosecution

0.78

 Investigations

0.78

 infiltration

0.75

 interviewing

0.74

 CASE

0.73

 Journalism

0.73

 Janeiro

0.72

CBI

0.71

Activations Density 0.621%

phrases related to investigations and probes conducted by authorities

phrases related to investigations and oversight processes

No Comments

No Known Activations