INDEX

Explanations

the word "Lincoln"

oai_token-act-pair · gpt-3.5-turbo

specific names or identifiers related to magnets

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

eele

-0.71

 needless

-0.67

KP

-0.63

 Laugh

-0.63

 comrade

-0.62

 sarc

-0.62

 misunderstand

-0.62

 misunderstanding

-0.62

 colleague

-0.62

 Mountain

-0.61

POSITIVE LOGITS

Ŀ

1.55

oln

1.06

¦

0.86

ħĭ

0.81

ypes

0.78

ģ

0.76

nen

0.74

zzle

0.73

ī

0.72

esta

0.70

Activations Density 0.000%

the word "Lincoln"

specific names or identifiers related to magnets

No Comments

No Known Activations