INDEX

Explanations

terms that have been mentioned or referenced previously in the text

oai_token-act-pair · gpt-3.5-turbo

the term "aforementioned" and related references to previously discussed topics or items

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ared

-0.80

asket

-0.72

aling

-0.71

airs

-0.70

elling

-0.68

ribution

-0.68

 Pike

-0.68

quest

-0.64

usk

-0.64

MET

-0.64

POSITIVE LOGITS

 aforementioned

0.90

 mentioned

0.79

mentioned

0.78

querque

0.73

 Deity

0.70

ãĥ´ãĤ¡

0.70

aurus

0.66

 disclaimer

0.65

ãĥ¤

0.64

 mention

0.63

Activations Density 0.008%

terms that have been mentioned or referenced previously in the text

the term "aforementioned" and related references to previously discussed topics or items

No Comments

No Known Activations