INDEX

Explanations

words related to investigations and politics

oai_token-act-pair · gpt-3.5-turbo

acronyms or abbreviations related to organizations or political entities

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

 trusted

-0.72

JFK

-0.69

 trademark

-0.68

 corrective

-0.66

 STATS

-0.64

gotten

-0.62

 corpor

-0.60

 quarantine

-0.60

 reward

-0.60

 Franchise

-0.59

POSITIVE LOGITS

culosis

0.93

henko

0.92

oshenko

0.85

chini

0.80

ulo

0.78

atoes

0.78

ucket

0.77

urgy

0.77

zx

0.77

emouth

0.76

Activations Density 0.230%

words related to investigations and politics

acronyms or abbreviations related to organizations or political entities

No Comments

No Known Activations