INDEX

Explanations

names containing "Anwar"

oai_token-act-pair · gpt-3.5-turbo

mentions of war-related terms

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

essee

-0.90

aminer

-0.87

sembly

-0.86

ĸļ

-0.80

choes

-0.74

ocre

-0.71

 suspic

-0.69

obook

-0.68

URES

-0.68

iquid

-0.66

POSITIVE LOGITS

rior

1.67

riors

1.56

fare

1.46

ring

0.99

lords

0.94

lord

0.93

locks

0.92

war

0.90

lock

0.88

ped

0.87

Activations Density 0.018%

names containing "Anwar"

mentions of war-related terms

No Comments

No Known Activations