INDEX

Explanations

descriptions or mentions of things that are simple

oai_token-act-pair · gpt-3.5-turbo

repeated references to the concept of simplicity

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

hovah

-0.77

vance

-0.74

hips

-0.74

 largeDownload

-0.73

 extensively

-0.73

raints

-0.68

 vigorously

-0.67

umar

-0.67

reon

-0.66

ibal

-0.66

POSITIVE LOGITS

tons

1.41

minded

1.01

 arithmetic

0.99

 syrup

0.97

ton

0.96

 minded

0.96

wallet

0.89

 pleasures

0.87

 sugars

0.84

 explanation

0.80

Activations Density 0.049%

descriptions or mentions of things that are simple

repeated references to the concept of simplicity

No Comments

No Known Activations