INDEX

Explanations

phrases related to seeking approval or support

oai_token-act-pair · gpt-3.5-turbo

expressions of political and social conflict or power dynamics

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

laughs

-0.81

 doesnt

-0.65

 refers

-0.63

laughter

-0.61

rontal

-0.60

 consists

-0.59

 Prosecutors

-0.58

 DOES

-0.57

 announces

-0.57

Alright

-0.57

POSITIVE LOGITS

 themselves

1.31

their

0.98

selves

0.95

 careers

0.90

 their

0.90

 theirs

0.85

 THEIR

0.80

 attest

0.79

 alike

0.77

 respectively

0.75

Activations Density 0.857%

phrases related to seeking approval or support

expressions of political and social conflict or power dynamics

No Comments

No Known Activations