INDEX

Explanations

mentions of closed spaces or objects

oai_token-act-pair · gpt-3.5-turbo

words related to closure or being closed

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

 generation

-0.67

 pieces

-0.64

mon

-0.64

iah

-0.63

 pile

-0.63

igans

-0.62

otaur

-0.61

 album

-0.61

 uprising

-0.60

 piece

-0.60

POSITIVE LOGITS

close

4.06

closed

2.78

Close

1.97

closure

1.85

CLOSE

1.28

 close

1.22

open

1.21

cl

1.14

older

1.13

 Close

1.10

Activations Density 0.014%

mentions of closed spaces or objects

words related to closure or being closed

No Comments

No Known Activations