INDEX

Explanations

the word "qual" with varying activation values

oai_token-act-pair · gpt-3.5-turbo

words that express quality or qualifications

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

orman

-1.01

eenth

-0.87

Construct

-0.81

ELD

-0.75

gerald

-0.72

wang

-0.71

esley

-0.71

 Rollins

-0.71

olph

-0.70

Gaming

-0.68

POSITIVE LOGITS

 sidx

1.01

 guiActiveUn

0.97

 qual

0.89

itatively

0.87

 externalToEVAOnly

0.86

 mathemat

0.82

icago

0.76

å£«

0.72

 ãĤµãĥ¼ãĥĨãĤ£ãĥ¯ãĥ³

0.71

 hiber

0.71

Activations Density 0.024%

the word "qual" with varying activation values

words that express quality or qualifications

No Comments

No Known Activations