INDEX

Explanations

words and phrases related to communication and interaction, particularly in online contexts

oai_token-act-pair · gpt-3.5-turbo

instances of affirmations or expressions of sentiment

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

Mobil

-0.64

 ÂŃ

-0.62

Individual

-0.61

Construction

-0.60

ATIVE

-0.60

Footnote

-0.59

 Fiscal

-0.57

Accept

-0.56

 Partnership

-0.56

 Thirty

-0.54

POSITIVE LOGITS

 dont

1.07

 english

1.03

 americ

1.00

 didnt

0.98

 whats

0.96

 david

0.96

 alot

0.95

 doesnt

0.95

 google

0.94

lol

0.92

Activations Density 0.951%

words and phrases related to communication and interaction, particularly in online contexts

instances of affirmations or expressions of sentiment

No Known Activations