INDEX

Explanations

words related to administrative and registration processes, such as "administered", "registries", and "laund..."

oai_token-act-pair · gpt-3.5-turbo

references to administrative and regulatory terminology

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

ãĥīãĥ©ãĤ´ãĥ³

-0.65

 Feast

-0.65

 behavi

-0.63

knife

-0.62

FUL

-0.61

 srfAttach

-0.61

ãģĦ

-0.60

FTWARE

-0.59

 phosphate

-0.58

 Canary

-0.58

POSITIVE LOGITS

ered

1.56

rator

1.40

rative

1.34

ration

1.26

ern

1.20

rators

1.19

ried

1.17

erers

1.16

rations

1.10

rial

1.09

Activations Density 0.041%

words related to administrative and registration processes, such as "administered", "registries", and "laund..."

references to administrative and regulatory terminology

No Comments

No Known Activations