INDEX

Explanations

mentions of bars or bar-related activities

oai_token-act-pair · gpt-3.5-turbo

mentions of bars

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

lihood

-0.97

UAL

-0.82

IBLE

-0.81

ãģį

-0.67

 Languages

-0.65

 Instruments

-0.65

 Sounders

-0.64

åĬ

-0.63

wip

-0.63

 Vehicles

-0.63

POSITIVE LOGITS

itone

1.37

celona

1.21

bell

1.12

becue

1.07

bara

1.07

bers

1.07

iatric

1.05

raged

1.02

ista

1.02

keep

0.97

Activations Density 0.031%

mentions of bars or bar-related activities

mentions of bars

No Comments

No Known Activations