INDEX

Explanations

Conjunctions followed by contradictions or alternative viewpoints

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

ãĤ¼ãĤ¦ãĤ¹

-0.97

 teasp

-0.94

ãĤ´

-0.87

BILITIES

-0.85

 pione

-0.84

Contents

-0.82

Ñı

-0.80

dayName

-0.80

č

-0.78

Ý

-0.78

POSITIVE LOGITS

 suppose

1.16

let

1.15

 consider

1.03

 suffice

1.01

 secondly

0.97

 surely

0.96

 therein

0.94

why

0.93

 alas

0.92

 imagine

0.92

Activations Density 0.345%

Conjunctions followed by contradictions or alternative viewpoints

No Comments

No Known Activations