INDEX

Explanations

computer code references or prompts for user inputs

oai_token-act-pair · gpt-3.5-turbo

sequences of characters or symbols, particularly those related to formatting or programming

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

wagon

-0.88

ciating

-0.82

stone

-0.82

rian

-0.80

ians

-0.79

icult

-0.79

oos

-0.78

lectic

-0.78

ular

-0.77

iac

-0.76

POSITIVE LOGITS

 Drac

0.78

 Dresden

0.75

 Laugh

0.68

 Higher

0.67

hler

0.67

âĸĪâĸĪâĸĪâĸĪâĸĪâĸĪâĸĪâĸĪ

0.65

ongyang

0.64

 Baltic

0.64

sembly

0.63

Wem

0.63

Activations Density 0.033%

computer code references or prompts for user inputs

sequences of characters or symbols, particularly those related to formatting or programming

No Comments

No Known Activations