INDEX

Explanations

words related to testing, trying out, or exploring different options and ideas

oai_token-act-pair · gpt-3.5-turbo

terms related to experimentation and testing

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 Cheong

-0.75

mary

-0.68

CLOSE

-0.66

games

-0.66

olulu

-0.65

ens

-0.64

Calling

-0.64

die

-0.64

si

-0.63

BN

-0.62

POSITIVE LOGITS

 experimentation

1.26

 experimenting

1.21

 experimented

1.13

 experiments

0.94

 tink

0.88

imental

0.88

 withd

0.86

 experiment

0.86

aults

0.85

 Experiment

0.81

Activations Density 0.007%

words related to testing, trying out, or exploring different options and ideas

terms related to experimentation and testing

No Comments

No Known Activations