INDEX

Explanations

phrases related to sports achievements or titles

oai_token-act-pair · gpt-3.5-turbo

references to various types of champions and championship titles

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

alle

-0.87

utterstock

-0.73

armac

-0.71

ument

-0.70

rams

-0.69

kinson

-0.69

oÄŁ

-0.69

lez

-0.69

esta

-0.67

okin

-0.66

POSITIVE LOGITS

Joined

0.79

 Steph

0.77

 Akira

0.72

 Julius

0.72

iatus

0.71

 Cyrus

0.68

Bj

0.68

 Connie

0.68

 Winston

0.68

 Jacques

0.68

Activations Density 0.081%

phrases related to sports achievements or titles

references to various types of champions and championship titles

No Comments

No Known Activations