INDEX

Explanations

phrases indicating a call to action or instruction, directing the reader to another location or specific action

oai_token-act-pair · gpt-3.5-turbo

references to navigating or accessing resources and links

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 hurled

-0.77

 conveyed

-0.72

 flung

-0.68

 employed

-0.68

iating

-0.65

iated

-0.64

 salute

-0.63

 blanket

-0.63

 struck

-0.62

 obliged

-0.62

POSITIVE LOGITS

www

1.15

 Settings

1.04

 Preferences

0.91

 Downloads

0.89

 http

0.81

 Menu

0.78

 STATS

0.77

 Github

0.74

 Finder

0.74

 Gaal

0.74

Activations Density 0.126%

phrases indicating a call to action or instruction, directing the reader to another location or specific action

references to navigating or accessing resources and links

No Comments

No Known Activations