INDEX

Explanations

phrases related to events, organizations, or individuals

oai_token-act-pair · gpt-3.5-turbo

titles and names of organizations and events

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ivably

-0.71

requires

-0.71

=]

-0.70

ourning

-0.68

]=

-0.68

should

-0.67

shall

-0.67

verbs

-0.66

iola

-0.65

|--

-0.65

POSITIVE LOGITS

 initiative

1.18

 project

1.10

 consortium

1.09

 program

1.07

 programme

1.04

 Foundation

1.03

 Initiative

1.01

 moniker

1.00

 debacle

0.97

 fiasco

0.95

Activations Density 0.520%

phrases related to events, organizations, or individuals

titles and names of organizations and events

No Comments

No Known Activations