INDEX

Explanations

numeric or random characters amidst text, possibly indicating a decoding or formatting anomaly

oai_token-act-pair · gpt-3.5-turbo

unique or special characters

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 heater

-0.64

 dissatisf

-0.63

 TABLE

-0.63

 repertoire

-0.62

eer

-0.62

eers

-0.60

rained

-0.59

 transient

-0.59

 Carney

-0.59

 Twain

-0.58

POSITIVE LOGITS

enberg

0.79

ukong

0.79

agram

0.77

imoto

0.72

onal

0.72

iversal

0.72

idon

0.71

ylon

0.70

uese

0.69

 watches

0.67

Activations Density 0.000%

numeric or random characters amidst text, possibly indicating a decoding or formatting anomaly

unique or special characters

No Comments

No Known Activations