INDEX

Explanations

phrases related to news and events

oai_token-act-pair · gpt-3.5-turbo

references to news sources and reported claims

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

ãĤª

-0.62

Maps

-0.54

 modern

-0.54

 whisky

-0.54

akespeare

-0.52

 feather

-0.52

ometimes

-0.52

 Publisher

-0.52

 Gael

-0.52

anguage

-0.51

POSITIVE LOGITS

 debacle

1.01

 fiasco

0.88

 ordeal

0.84

 incident

0.78

 spree

0.74

emate

0.68

 saga

0.68

 predicament

0.67

stros

0.67

 case

0.63

Activations Density 0.894%

phrases related to news and events

references to news sources and reported claims

No Comments

No Known Activations