INDEX

Explanations

words and phrases related to strong or extreme emotions or qualities

oai_token-act-pair · gpt-3.5-turbo

instances of numerical representations or ratings in text

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

inav

-0.66

 mosqu

-0.66

 commod

-0.65

 escort

-0.65

 fame

-0.65

 shelters

-0.65

 residence

-0.64

 nearest

-0.64

 handy

-0.63

 assemb

-0.63

POSITIVE LOGITS

yet

0.98

unless

0.97

although

0.96

ï¸ı

0.94

since

0.91

though

0.87

especially

0.87

nor

0.87

because

0.87

albeit

0.86

Activations Density 0.329%

words and phrases related to strong or extreme emotions or qualities

instances of numerical representations or ratings in text

No Comments

No Known Activations