INDEX

Explanations

statements indicating concern or impact on a larger scale

oai_token-act-pair · gpt-3.5-turbo

statements about the status or condition of various situations and issues

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 inquire

-0.72

 WRITE

-0.68

 fishes

-0.66

eva

-0.64

 Acqu

-0.63

CHO

-0.63

 shuffle

-0.62

itely

-0.62

 confidentiality

-0.61

 Choose

-0.61

POSITIVE LOGITS

 causing

1.20

 sympt

1.18

 fueling

1.15

 harming

1.14

fue

1.14

 exacerb

1.13

 undermining

1.12

 overshadow

1.10

 prompting

1.07

 impacting

1.06

Activations Density 0.197%

statements indicating concern or impact on a larger scale

statements about the status or condition of various situations and issues

No Comments

No Known Activations