INDEX

Explanations

online-related terms or references

oai_token-act-pair · gpt-3.5-turbo

references to online platforms and communities

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 marsh

-0.74

 compens

-0.70

 lull

-0.64

 mustard

-0.64

 fort

-0.63

 detach

-0.62

 glim

-0.62

 decre

-0.61

 bott

-0.60

 shepherd

-0.59

POSITIVE LOGITS

 Online

4.07

Online

2.76

online

2.31

 online

2.17

 Offline

1.73

 offline

1.45

Offline

1.32

 Interactive

1.32

 Internet

1.27

Internet

1.22

Activations Density 0.015%

online-related terms or references

references to online platforms and communities

No Comments

No Known Activations