INDEX

Explanations

phrases related to answers or responses, particularly with emphasis

oai_token-act-pair · gpt-3.5-turbo

statements regarding answers to yes/no questions

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

models

-0.68

Person

-0.67

astical

-0.67

 Franch

-0.66

ventures

-0.66

usp

-0.66

vre

-0.66

sers

-0.65

jan

-0.64

rongh

-0.63

POSITIVE LOGITS

yes

1.32

YES

1.17

 unequiv

1.12

 affirmative

1.04

 simple

1.00

 emph

0.98

 obvious

0.97

YES

0.95

 straightforward

0.89

nil

0.87

Activations Density 0.107%

phrases related to answers or responses, particularly with emphasis

statements regarding answers to yes/no questions

No Comments

No Known Activations