INDEX

Explanations

adjectives indicating a preference or goal

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "desired" 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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

load

-0.66

 Methodist

-0.64

Medic

-0.61

punk

-0.60

sha

-0.59

inosaur

-0.58

zens

-0.57

 idiots

-0.57

 Anyway

-0.56

inc

-0.55

POSITIVE LOGITS

 desired

3.95

 desirable

2.06

 preferred

1.88

 wished

1.50

 undesirable

1.49

 coveted

1.49

 requested

1.48

 dreaded

1.39

 needed

1.38

 predetermined

1.36

Activations Density 0.011%

adjectives indicating a preference or goal

occurrences of the word "desired" in various contexts

No Comments

No Known Activations