INDEX

Explanations

phrases describing things that are located or happening on the outside

oai_token-act-pair · gpt-3.5-turbo

references to external locations or contexts

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

 commission

-0.62

 coll

-0.61

 leve

-0.60

uly

-0.58

 liar

-0.57

 quadru

-0.57

 thanks

-0.55

 foll

-0.55

 syndrome

-0.55

eries

-0.55

POSITIVE LOGITS

Outside

3.50

 Outside

3.19

outside

2.36

Inside

1.91

 Inside

1.85

 outside

1.71

inside

1.64

Elsewhere

1.33

 Across

1.24

Across

1.23

Activations Density 0.015%

phrases describing things that are located or happening on the outside

references to external locations or contexts

No Comments

No Known Activations