INDEX

Explanations

personal opinions or thoughts expressed in the first person

oai_token-act-pair · gpt-3.5-turbo

first-person pronouns indicating self-reference

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

ļéĨĴ

-0.66

affles

-0.61

screen

-0.58

ienne

-0.58

kward

-0.56

 Rockefeller

-0.56

 Bent

-0.55

 Papers

-0.55

 Jindal

-0.55

 Butter

-0.54

POSITIVE LOGITS

'm

1.48

've

1.35

'll

1.21

 guess

1.08

 suppose

1.07

'd

1.05

nex

0.97

 think

0.94

 believe

0.92

MAX

0.90

Activations Density 0.279%

personal opinions or thoughts expressed in the first person

first-person pronouns indicating self-reference

No Comments

No Known Activations