INDEX

Explanations

terms related to joining or participating in a group or activity

oai_token-act-pair · gpt-3.5-turbo

gerunds and participles

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

cop

-0.63

dh

-0.61

 beta

-0.60

 plaus

-0.59

 invari

-0.59

Om

-0.58

 Reach

-0.58

 compatible

-0.58

Ty

-0.57

CB

-0.57

POSITIVE LOGITS

ining

4.30

ined

2.96

iners

2.20

iner

2.04

ines

2.00

inery

1.77

inition

1.73

ine

1.68

inement

1.57

ineries

1.56

Activations Density 0.005%

terms related to joining or participating in a group or activity

gerunds and participles

No Comments

No Known Activations