INDEX

Explanations

adjective-noun pairs emphasizing extremity or completeness

oai_token-act-pair · gpt-3.5-turbo

phrases indicating totality or completeness in various contexts

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

IFrame

Link

Negative Logits

swick

-1.02

RPG

-0.89

EStream

-0.81

roups

-0.74

ORPG

-0.73

 RPGs

-0.72

Kids

-0.72

bees

-0.71

yles

-0.69

ESA

-0.69

POSITIVE LOGITS

 unmist

1.05

 unamb

1.04

 emph

1.03

 unequiv

1.03

rogen

1.02

 undeniable

0.99

 unconditional

0.94

 conclusive

0.90

 coherent

0.90

 decisive

0.89

Activations Density 0.185%

adjective-noun pairs emphasizing extremity or completeness

phrases indicating totality or completeness in various contexts

No Known Activations