INDEX

Explanations

phrases indicating that something is not yet decided, confirmed, or known

oai_token-act-pair · gpt-3.5-turbo

phrases related to the status of pending actions or confirmations

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Guilty

-0.80

 omit

-0.71

 srfAttach

-0.71

 Falling

-0.68

 complicity

-0.67

 Lose

-0.66

 fail

-0.66

Fail

-0.65

 disobedience

-0.65

 guilty

-0.65

POSITIVE LOGITS

 ascert

1.33

 disclosed

1.23

 negotiated

1.20

 explored

1.19

 finalized

1.19

 vetted

1.16

 clarified

1.10

 resolved

1.09

 communicated

1.09

 revealed

1.08

Activations Density 0.144%

phrases indicating that something is not yet decided, confirmed, or known

phrases related to the status of pending actions or confirmations

No Comments

No Known Activations