INDEX

Explanations

instances where someone's actions are contrasted with a subsequent outcome or decision

oai_token-act-pair · gpt-3.5-turbo

phrases that introduce contrast or exceptions in statements

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

arity

-0.73

uto

-0.71

edu

-0.70

MI

-0.68

urg

-0.68

MO

-0.65

irt

-0.65

oir

-0.65

mast

-0.65

irm

-0.64

POSITIVE LOGITS

 alas

1.10

 luckily

1.00

 fortunately

0.98

 succumbed

0.93

 thankfully

0.92

 unfortunately

0.92

 lacked

0.92

 lacks

0.89

 struggled

0.87

 opted

0.87

Activations Density 0.162%

instances where someone's actions are contrasted with a subsequent outcome or decision

phrases that introduce contrast or exceptions in statements

No Comments

No Known Activations