INDEX

Explanations

phrases indicating a lack of connection or association between different concepts

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate a disconnection or lack of relevance to a specific topic or issue

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

©¶æ¥µ

-0.78

 Ahead

-0.72

NN

-0.68

ura

-0.68

warn

-0.68

kept

-0.67

emies

-0.66

bryce

-0.64

Released

-0.64

Used

-0.63

POSITIVE LOGITS

 regard

0.80

erous

0.76

 regards

0.73

 what

0.68

 impunity

0.67

 fiat

0.65

 nickel

0.62

rament

0.61

 entitle

0.60

 money

0.60

Activations Density 0.041%

phrases indicating a lack of connection or association between different concepts

phrases that indicate a disconnection or lack of relevance to a specific topic or issue

No Comments

No Known Activations