INDEX

Explanations

expressions confirming a correctness or accuracy

oai_token-act-pair · gpt-3.5-turbo

affirmations of correctness or agreement

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

ains

-0.64

neys

-0.61

utters

-0.61

craft

-0.59

graph

-0.58

otton

-0.57

kel

-0.56

ility

-0.56

%]

-0.55

legram

-0.55

POSITIVE LOGITS

eous

1.08

terday

0.84

wing

0.76

 fielder

0.74

 winger

0.72

footed

0.72

 wing

0.72

 aligned

0.71

fully

0.69

itudinal

0.67

Activations Density 0.035%

expressions confirming a correctness or accuracy

affirmations of correctness or agreement

No Comments

No Known Activations