INDEX

Explanations

direct responses or answers to questions or requests

oai_token-act-pair · gpt-3.5-turbo

instances of answering or responding to questions

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

 survived

-0.71

 "$:/

-0.71

rawdownloadcloneembedreportprint

-0.67

 compensated

-0.64

 Surv

-0.62

 implements

-0.62

Tools

-0.61

heid

-0.61

bilt

-0.61

 Signed

-0.61

POSITIVE LOGITS

 question

1.09

 questions

1.02

 criticisms

1.00

 queries

0.97

 request

0.95

 requests

0.93

 inquiries

0.91

 questionnaire

0.89

 summons

0.89

 query

0.88

Activations Density 0.236%

direct responses or answers to questions or requests

instances of answering or responding to questions

No Comments

No Known Activations