INDEX

Explanations

terms related to specific names and technical terms such as company names, system codes, and objects

oai_token-act-pair · gpt-3.5-turbo

specific names or terms associated with various subjects or entities

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

lihood

-0.65

taboola

-0.61

 HuffPost

-0.57

 ..........

-0.57

 âĢº

-0.54

tesy

-0.52

----------

-0.50

nos

-0.49

 largeDownload

-0.49

-------------

-0.49

POSITIVE LOGITS

rul

0.71

idious

0.66

 shenan

0.65

eteria

0.61

 Citiz

0.60

 lineback

0.59

unia

0.59

 destro

0.58

'."

0.58

 muse

0.57

Activations Density 2.394%

terms related to specific names and technical terms such as company names, system codes, and objects

specific names or terms associated with various subjects or entities

No Known Activations