INDEX

Explanations

instances of the word "lime"

oai_token-act-pair · gpt-3.5-turbo

references to lime and its various associations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

bucks

-0.78

rien

-0.74

mat

-0.73

geons

-0.72

owe

-0.71

typ

-0.71

caster

-0.70

ailed

-0.69

 externalToEVAOnly

-0.68

apter

-0.68

POSITIVE LOGITS

 Lime

0.86

Pie

0.81

vous

0.76

 lime

0.74

 Cout

0.74

ez

0.68

lime

0.67

 Juice

0.67

lli

0.67

Ã±

0.67

Activations Density 0.029%

instances of the word "lime"

references to lime and its various associations

No Comments

No Known Activations