INDEX

Explanations

answers to questions

oai_token-act-pair · gpt-3.5-turbo

references to "answers" in various contexts

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

base

-0.67

vest

-0.63

vert

-0.63

 spree

-0.60

erial

-0.60

strip

-0.59

station

-0.59

tern

-0.59

ITAL

-0.59

porary

-0.58

POSITIVE LOGITS

 answers

4.06

 answer

2.70

 Answers

2.62

 answered

2.25

 replies

2.19

answer

2.10

 Answer

2.06

 answering

2.03

Answer

2.00

swers

1.99

Activations Density 0.006%

answers to questions

references to "answers" in various contexts

No Comments

No Known Activations