INDEX

Explanations

the word "kind" with a high level of activation

oai_token-act-pair · gpt-3.5-turbo

phrases referring to different categories or types of things

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

 VIDEOS

-0.79

UNCH

-0.74

mercial

-0.70

å§«

-0.70

oulos

-0.68

è¦ļéĨĴ

-0.68

eor

-0.67

edia

-0.66

 Minutes

-0.66

borough

-0.65

POSITIVE LOGITS

lier

0.89

hearted

0.84

liest

0.82

liness

0.78

ifier

0.75

ling

0.74

 gesture

0.72

ilege

0.72

 prevail

0.68

nered

0.66

Activations Density 0.031%

the word "kind" with a high level of activation

phrases referring to different categories or types of things

No Comments

No Known Activations