INDEX

Explanations

the letter 'h'

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

words with some variations of the letter 'h' at a high activation level

oai_token-act-pair · gpt-3.5-turbo

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

ãĤ´ãĥ³

-0.89

ãĥ¯

-0.83

éĹĺ

-0.78

ãĥ¼ãĥĨãĤ£

-0.74

ãĥ´ãĤ¡

-0.71

EStream

-0.68

DragonMagazine

-0.68

 totality

-0.65

å§«

-0.64

enhagen

-0.64

POSITIVE LOGITS

oused

1.30

awk

1.22

ousing

1.19

ulk

1.15

idd

1.14

acking

1.14

anging

1.11

ashing

1.11

olly

1.10

anky

1.09

Activations Density 0.019%

the letter 'h'

words with some variations of the letter 'h' at a high activation level

No Comments

No Known Activations