INDEX

Explanations

characters or symbols commonly found in text rather than meaningful words or phrases

oai_token-act-pair · gpt-3.5-turbo

non-standard or non-English characters and symbols

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

anwhile

-0.82

ftime

-0.76

Beh

-0.70

theless

-0.70

estation

-0.68

espie

-0.68

 agre

-0.63

 Warn

-0.62

bro

-0.62

 Brennan

-0.61

POSITIVE LOGITS

¥

2.00

º

1.94

´

1.93

²

1.91

½

1.91

»

1.91

¨

1.91

±

1.90

ı

1.90

İ

1.89

Activations Density 0.015%

characters or symbols commonly found in text rather than meaningful words or phrases

non-standard or non-English characters and symbols

No Comments

No Known Activations