INDEX

Explanations

phrases related to preventing or regretting negative events

oai_token-act-pair · gpt-3.5-turbo

phrases related to preventing undesirable events or situations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

Reviewer

-0.98

 searching

-0.76

 stocking

-0.68

 stout

-0.67

 touting

-0.66

 Search

-0.65

 Editors

-0.63

 Recomm

-0.63

 offering

-0.62

 composing

-0.61

POSITIVE LOGITS

 happen

2.19

 happened

2.05

 occur

2.04

 happening

2.01

 occurred

1.94

 occurs

1.88

 happens

1.88

 occurring

1.85

 happ

1.74

 transpired

1.61

Activations Density 0.587%

phrases related to preventing or regretting negative events

phrases related to preventing undesirable events or situations

No Comments

No Known Activations