INDEX

Explanations

instances of the word "wives."

oai_token-act-pair · gpt-3.5-turbo

references to clothing and related terms

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

 Glacier

-0.86

 Albion

-0.70

 horizont

-0.67

LLP

-0.65

 trunc

-0.63

 swarm

-0.63

 spectral

-0.63

 decomp

-0.62

 fragmentation

-0.62

EMENT

-0.61

POSITIVE LOGITS

othes

2.67

otle

2.60

oan

1.70

oling

1.47

ophe

1.36

othe

1.36

utra

1.33

othing

1.32

oled

1.31

 wives

1.19

Activations Density 0.039%

instances of the word "wives."

references to clothing and related terms

No Comments

No Known Activations