INDEX

Explanations

phrases related to being confined or restricted in some way

oai_token-act-pair · gpt-3.5-turbo

references to being confined or restricted

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

issance

-1.00

resso

-0.79

lus

-0.75

ciation

-0.74

annis

-0.74

lication

-0.70

lene

-0.70

arenthood

-0.69

enegger

-0.69

 Lynd

-0.67

POSITIVE LOGITS

door

0.82

 doors

0.80

 locked

0.79

picking

0.76

 onto

0.74

locked

0.74

 shut

0.73

doors

0.72

 chests

0.71

 door

0.69

Activations Density 0.031%

phrases related to being confined or restricted in some way

references to being confined or restricted

No Comments

No Known Activations