INDEX

Explanations

verbs corresponding to imperative or future intent

oai_token-act-pair · gpt-3.5-turbo

forms of the verb "be."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

izable

-0.80

stuff

-0.79

 separat

-0.72

 uniqueness

-0.71

 cubes

-0.71

 Stuff

-0.70

 validity

-0.68

ipeg

-0.66

 immortality

-0.66

 deletion

-0.66

POSITIVE LOGITS

 aware

1.15

 reluctant

1.12

 able

1.11

 interested

1.08

 concerned

1.08

 obliged

1.05

 negligent

1.05

 alerted

1.03

 unable

1.02

 obligated

1.00

Activations Density 0.446%

verbs corresponding to imperative or future intent

forms of the verb "be."

No Comments

No Known Activations