INDEX

Explanations

the word "Mirror" followed by a number from 9 to 10

oai_token-act-pair · gpt-3.5-turbo

mentions of the term "Mirror" 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 @ 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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ourke

-0.73

yright

-0.70

reat

-0.67

ales

-0.67

reek

-0.67

ential

-0.66

athetic

-0.65

RAFT

-0.65

adding

-0.64

akings

-0.64

POSITIVE LOGITS

 Mirror

1.29

ror

0.88

 istg

0.80

 mirror

0.76

 Scroll

0.74

 Divinity

0.71

Mir

0.70

 discrep

0.70

wip

0.69

angelo

0.67

Activations Density 0.005%

the word "Mirror" followed by a number from 9 to 10

mentions of the term "Mirror" in various contexts

No Comments

No Known Activations