INDEX

Explanations

mentions or references to the word "tin"

oai_token-act-pair · gpt-3.5-turbo

references to tin

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

INAL

-0.74

 SYSTEM

-0.74

IST

-0.71

Variable

-0.68

reviewed

-0.67

AE

-0.67

 Represent

-0.63

 Accountability

-0.63

 Vector

-0.63

EM

-0.62

POSITIVE LOGITS

kered

1.11

ned

1.01

tin

1.01

oshenko

1.00

uous

0.89

icum

0.87

ween

0.87

querque

0.86

 oxide

0.86

oco

0.85

Activations Density 0.007%

mentions or references to the word "tin"

references to tin

No Comments

No Known Activations