INDEX

Explanations

descriptive words or phrases related to various concepts and ideas

oai_token-act-pair · gpt-3.5-turbo

references to humor, social commentary, and pop culture concepts

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

Ĭ±

-0.72

nces

-0.71

Comments

-0.69

 Latest

-0.69

itudes

-0.67

urations

-0.67

ousands

-0.66

azes

-0.66

ICES

-0.66

tails

-0.65

POSITIVE LOGITS

 unto

0.78

 breaker

0.75

 ploy

0.72

brainer

0.71

 whore

0.71

 affair

0.70

 deterrent

0.70

 thing

0.69

 puzz

0.69

 staple

0.68

Activations Density 0.688%

descriptive words or phrases related to various concepts and ideas

references to humor, social commentary, and pop culture concepts

No Comments

No Known Activations