INDEX

Explanations

terms related to computer hardware or software

oai_token-act-pair · gpt-3.5-turbo

references to various types of software or products labeled as "ware."

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

 Albany

-0.69

 trib

-0.65

 Reneg

-0.61

 intro

-0.60

 alternate

-0.60

 Extra

-0.60

 Rachel

-0.60

 varied

-0.59

 improv

-0.59

 interviews

-0.59

POSITIVE LOGITS

ware

5.02

WARE

2.04

Ware

1.82

 ware

1.81

eware

1.41

Software

1.20

wear

1.12

washer

1.01

software

1.01

aware

1.00

Activations Density 0.022%

terms related to computer hardware or software

references to various types of software or products labeled as "ware."

No Comments

No Known Activations