INDEX

Explanations

phrases related to news articles, investigations, and official reports

oai_token-act-pair · gpt-3.5-turbo

punctuations indicating the end of sentences or statements

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

 pudding

-0.76

 fists

-0.74

 extinct

-0.73

 imperson

-0.72

 backward

-0.72

 impression

-0.71

 tiger

-0.70

 hands

-0.70

 utter

-0.69

 purse

-0.69

POSITIVE LOGITS

sbm

0.95

 Similarly

0.93

 Additionally

0.91

 Flavoring

0.90

 Previously

0.89

 Specifically

0.88

 Furthermore

0.88

 Along

0.88

 Also

0.87

 Moreover

0.87

Activations Density 0.253%

phrases related to news articles, investigations, and official reports

punctuations indicating the end of sentences or statements

No Comments

No Known Activations