INDEX

Explanations

links or instructions for more information

oai_token-act-pair · gpt-3.5-turbo

phrases prompting the reader to click or check out additional content or 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

vernment

-0.86

 recovers

-0.69

 nurs

-0.67

 healed

-0.65

 rebound

-0.61

 restores

-0.61

 condem

-0.61

 bluff

-0.60

ciating

-0.58

 precision

-0.57

POSITIVE LOGITS

ravis

0.74

rss

0.72

 subscribe

0.71

ULE

0.67

leases

0.64

 Forbes

0.62

gur

0.62

 Subscribe

0.62

Visit

0.61

yip

0.61

Activations Density 0.108%

links or instructions for more information

phrases prompting the reader to click or check out additional content or links

No Comments

No Known Activations