INDEX

Explanations

terms related to achieving peak performance or reaching maximum levels

oai_token-act-pair · gpt-3.5-turbo

terms related to highs and lows in various 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

DI

-0.74

 subclass

-0.71

armed

-0.69

rib

-0.64

edia

-0.64

page

-0.63

 Wing

-0.62

va

-0.62

 automatically

-0.62

cha

-0.59

POSITIVE LOGITS

 highs

4.24

 lows

3.41

 peaks

1.79

 heights

1.45

 positives

1.35

ups

1.27

 extremes

1.22

 dips

1.21

 ceilings

1.17

 milestones

1.16

Activations Density 0.014%

terms related to achieving peak performance or reaching maximum levels

terms related to highs and lows in various contexts

No Comments

No Known Activations