INDEX

Explanations

specific years in the text

oai_token-act-pair · gpt-3.5-turbo

references to the year 2013

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

hyde

-0.71

 stret

-0.71

othe

-0.64

tons

-0.64

Pg

-0.64

çļ

-0.64

 sympath

-0.64

 ether

-0.63

 BahÃ¡

-0.62

 inclined

-0.61

POSITIVE LOGITS

3.30

2.86

2.85

2.71

2.70

2.59

2.51

2.50

2.37

2.33

Activations Density 0.021%

specific years in the text

references to the year 2013

No Comments

No Known Activations