INDEX

Explanations

numerical values that are specifically nine or ten

oai_token-act-pair · gpt-3.5-turbo

phrases indicating the concept of "one" or singularity

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

ories

-0.76

igion

-0.70

 launchers

-0.67

avascript

-0.67

ONSORED

-0.65

izations

-0.64

utterstock

-0.64

gy

-0.63

 Shutterstock

-0.63

 Modes

-0.63

POSITIVE LOGITS

 dimensional

1.02

 sided

0.89

dimensional

0.82

 hundred

0.82

âĢĳ

0.79

 Thousand

0.76

 Hundred

0.75

xious

0.72

 thousand

0.71

cott

0.71

Activations Density 0.026%

numerical values that are specifically nine or ten

phrases indicating the concept of "one" or singularity

No Comments

No Known Activations