INDEX

Explanations

the word "mit" or variations of it in the texts

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "mit."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

Ļ½

-0.68

 tremend

-0.67

¥ŀ

-0.65

ighth

-0.61

 Borders

-0.60

 pharmacies

-0.60

spr

-0.59

lets

-0.59

hospital

-0.58

 grid

-0.58

POSITIVE LOGITS

mit

1.21

chell

1.19

ochond

1.05

igation

1.01

igated

0.92

ropolitan

0.88

rique

0.88

igating

0.86

ulous

0.85

ulously

0.83

Activations Density 0.013%

the word "mit" or variations of it in the texts

occurrences of the word "mit."

No Comments

No Known Activations