INDEX

Explanations

phrases indicating an alternative or different option

oai_token-act-pair · gpt-3.5-turbo

references to the company Elsevier and its publications

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

 intercept

-0.74

 perfect

-0.68

 rods

-0.65

twitch

-0.62

 prom

-0.62

voy

-0.61

reenshot

-0.60

 ribbon

-0.60

API

-0.60

mir

-0.59

POSITIVE LOGITS

 Else

3.73

Else

2.35

grave

1.67

 Springer

1.57

 Otherwise

1.47

 Wiley

1.27

Elsewhere

1.25

else

1.08

Otherwise

1.06

urther

1.02

Activations Density 0.039%

phrases indicating an alternative or different option

references to the company Elsevier and its publications

No Comments

No Known Activations