INDEX

Explanations

phrases emphasizing contrast or preference

oai_token-act-pair · gpt-3.5-turbo

phrases that express negation or contrastive ideas

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

 Norn

-0.68

nec

-0.67

omatic

-0.64

CLASSIFIED

-0.64

 clearance

-0.63

CF

-0.63

ced

-0.62

}}}

-0.62

cia

-0.62

 Grounds

-0.61

POSITIVE LOGITS

 reinvent

0.93

 rely

0.87

 blindly

0.86

 succumb

0.85

 relying

0.84

 necessarily

0.84

 simply

0.83

 merely

0.83

 speculate

0.81

rahim

0.78

Activations Density 0.156%

phrases emphasizing contrast or preference

phrases that express negation or contrastive ideas

No Comments

No Known Activations