INDEX

Explanations

phrases indicating alignment or agreement

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate alignment or agreement with policies or standards

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

©¶æ¥µ

-0.86

chief

-0.77

quer

-0.77

hops

-0.73

orah

-0.73

Chat

-0.71

chat

-0.71

agg

-0.71

raq

-0.64

borgh

-0.64

POSITIVE LOGITS

 regards

1.12

 regard

1.10

 respect

0.97

 impunity

0.87

stood

0.86

draw

0.82

rium

0.80

drawn

0.75

 intact

0.71

 dignity

0.69

Activations Density 0.111%

phrases indicating alignment or agreement

phrases that indicate alignment or agreement with policies or standards

No Comments

No Known Activations