INDEX

Explanations

adjectives or verbs indicating a change in ease or difficulty of a certain action

oai_token-act-pair · gpt-3.5-turbo

phrases indicating the relative difficulty or ease of various actions or regulations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

notations

-0.65

Originally

-0.63

ilo

-0.63

leigh

-0.60

agraph

-0.60

milo

-0.60

Blu

-0.58

 Difference

-0.57

wings

-0.56

kind

-0.56

POSITIVE LOGITS

for

0.79

to

0.78

 punishable

0.71

IBLE

0.67

ible

0.65

enged

0.64

 prey

0.63

for

0.63

 easier

0.62

anced

0.62

Activations Density 0.076%

adjectives or verbs indicating a change in ease or difficulty of a certain action

phrases indicating the relative difficulty or ease of various actions or regulations

No Comments

No Known Activations