INDEX

Explanations

the word "essentially" with high activation values

oai_token-act-pair · gpt-3.5-turbo

the word "essentially" and its variants, indicating a focus on defining or clarifying concepts

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

Ey

-0.68

"},{"

-0.64

Known

-0.60

ft

-0.60

-0.59

 Passion

-0.59

fal

-0.58

icy

-0.58

ieth

-0.57

POSITIVE LOGITS

 essentially

3.05

 basically

2.62

 effectively

1.82

 fundamentally

1.70

 ostensibly

1.63

 virtually

1.54

 literally

1.49

 largely

1.48

 practically

1.46

 Essentially

1.44

Activations Density 0.013%

the word "essentially" with high activation values

the word "essentially" and its variants, indicating a focus on defining or clarifying concepts

No Comments

No Known Activations