INDEX

Explanations

phrases related to making attempts or efforts

oai_token-act-pair · gpt-3.5-turbo

instances of the word "tries."

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

 Matter

-0.64

 insign

-0.61

âĢİ

-0.59

 entitle

-0.57

icken

-0.57

 cloud

-0.56

amen

-0.55

 grave

-0.55

 Columb

-0.54

 subject

-0.54

POSITIVE LOGITS

 tries

3.27

 attempts

1.95

try

1.72

 strives

1.72

 seeks

1.57

try

1.53

 tried

1.52

 listens

1.41

Try

1.39

 succeeds

1.38

Activations Density 0.013%

phrases related to making attempts or efforts

instances of the word "tries."

No Comments

No Known Activations