INDEX

Explanations

phrases related to time and specifics about numbers or measurements

oai_token-act-pair · gpt-3.5-turbo

references to cyclic or iterative concepts

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Ã§

-0.65

acca

-0.64

iments

-0.61

 Bundes

-0.61

 Mehran

-0.61

 Motorsport

-0.59

enez

-0.58

ãĤ¿

-0.57

 Strongh

-0.57

yer

-0.56

POSITIVE LOGITS

table

1.47

about

1.46

trip

1.40

abouts

1.34

rob

1.11

worms

1.10

tri

1.10

worm

1.07

house

1.03

ups

1.02

Activations Density 0.026%

phrases related to time and specifics about numbers or measurements

references to cyclic or iterative concepts

No Comments

No Known Activations