INDEX

Explanations

phrases indicating a suggestion or advice to take action

oai_token-act-pair · gpt-3.5-turbo

expressions emphasizing the need for improvement or preparation

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

 Pione

-0.76

icidal

-0.71

 Indigo

-0.64

 Pike

-0.63

 Pavilion

-0.63

 Emer

-0.62

oidal

-0.62

Nob

-0.62

 Railroad

-0.60

 Nord

-0.60

POSITIVE LOGITS

 than

0.91

 suited

0.91

 behaved

0.81

than

0.78

 cannabin

0.73

 Than

0.72

Ĥİ

0.70

seller

0.69

 acquainted

0.68

 quantify

0.68

Activations Density 0.030%

phrases indicating a suggestion or advice to take action

expressions emphasizing the need for improvement or preparation

No Comments

No Known Activations