INDEX

Explanations

profile-related information, such as user profiles and percentages

oai_token-act-pair · gpt-3.5-turbo

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 occas

-0.80

 reluct

-0.78

quartered

-0.72

 enthusi

-0.72

 favourable

-0.72

 tremend

-0.70

 toget

-0.68

bom

-0.68

cffff

-0.68

 conflic

-0.67

POSITIVE LOGITS

 Presents

0.80

Mk

0.73

Â®

0.68

 Mahjong

0.67

yll

0.64

mk

0.64

Sys

0.64

 Solitaire

0.62

\":

0.62

ļéĨĴ

0.62

Activations Density 0.098%

profile-related information, such as user profiles and percentages

No Comments

No Known Activations