INDEX

Explanations

proper nouns related to a specific name or concept ("ram")

oai_token-act-pair · gpt-3.5-turbo

references to the name "Eram."

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

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

 Yankee

-0.68

 Carib

-0.65

 predic

-0.61

 indifference

-0.61

 subp

-0.59

 pitcher

-0.59

LCS

-0.57

urses

-0.57

 Panthers

-0.55

bed

-0.54

POSITIVE LOGITS

irez

1.18

ming

1.11

bling

1.03

matical

1.02

mers

1.02

pton

1.01

bler

1.01

atic

0.97

med

0.96

bles

0.92

Activations Density 0.014%

proper nouns related to a specific name or concept ("ram")

references to the name "Eram."

No Comments

No Known Activations