INDEX

Explanations

mentions of words starting with "kn" or "ight" followed by other letters

oai_token-act-pair · gpt-3.5-turbo

references to specific instances or forms of the word "knif"

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

 endangered

-0.73

 showc

-0.70

ģ«

-0.70

catentry

-0.69

hetical

-0.68

ICAN

-0.67

tnc

-0.65

 deficient

-0.64

 overdue

-0.64

 impair

-0.63

POSITIVE LOGITS

pton

0.89

Kn

0.81

fold

0.78

bart

0.78

leaf

0.72

itzer

0.72

puff

0.71

kn

0.71

NK

0.71

edge

0.70

Activations Density 0.099%

mentions of words starting with "kn" or "ight" followed by other letters

references to specific instances or forms of the word "knif"

No Comments

No Known Activations