INDEX

Explanations

terms related to physical places or locations

oai_token-act-pair · gpt-3.5-turbo

words and concepts related to crime and danger

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

emale

-0.65

 Niet

-0.61

 Frie

-0.60

farious

-0.58

 helicop

-0.55

 Azerb

-0.55

 Nare

-0.55

ADRA

-0.55

lyak

-0.54

pheus

-0.54

POSITIVE LOGITS

 TEXT

0.59

lement

0.59

0.58

actionDate

0.57

 âĢº

0.54

 hoax

0.53

][

0.52

BUS

0.51

::

0.50

 Vulkan

0.50

Activations Density 0.329%

terms related to physical places or locations

words and concepts related to crime and danger

No Comments

No Known Activations