INDEX

Explanations

instances where something is described in a technical or legal manner

oai_token-act-pair · gpt-3.5-turbo

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

ieties

-0.79

Io

-0.69

ĸļ

-0.68

 Decay

-0.68

Vaj

-0.66

 Surge

-0.66

ials

-0.64

oday

-0.61

è¦ļéĨĴ

-0.61

orters

-0.60

POSITIVE LOGITS

 speaking

0.96

 prohibited

0.87

 obligated

0.83

 minded

0.82

 feasible

0.81

 barred

0.80

 correct

0.78

allowed

0.77

 forbidden

0.76

 allowed

0.75

Activations Density 0.081%

instances where something is described in a technical or legal manner

No Comments

No Known Activations