INDEX

Explanations

locations or places

oai_token-act-pair · gpt-3.5-turbo

phrases related to arrests and legal situations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

¥µ

-0.69

catentry

-0.67

He

-0.66

he

-0.66

ħĭ

-0.64

ython

-0.63

his

-0.62

Ô

-0.61

inav

-0.61

hes

-0.61

POSITIVE LOGITS

 respectively

0.84

Their

0.81

 Both

0.80

 Their

0.79

 Fleming

0.77

 Nichols

0.77

 Goodwin

0.74

 Payne

0.74

Ms

0.74

 trio

0.74

Activations Density 0.572%

locations or places

phrases related to arrests and legal situations

No Comments

No Known Activations