INDEX

Explanations

verbs related to directing or adhering to instructions or guidelines

oai_token-act-pair · gpt-3.5-turbo

instances of the word "follow."

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 wounding

-0.78

orc

-0.66

ukemia

-0.66

pite

-0.63

ldom

-0.62

aucas

-0.62

adin

-0.61

intendent

-0.61

morph

-0.61

obyl

-0.60

POSITIVE LOGITS

follow

1.09

 follow

1.04

 follows

0.94

 Follow

0.94

LLOW

0.81

 followed

0.80

SHIP

0.77

ansen

0.72

ĸļ

0.69

Follow

0.67

Activations Density 0.012%

verbs related to directing or adhering to instructions or guidelines

instances of the word "follow."

No Comments

No Known Activations