INDEX

Explanations

statements followed by a negation

oai_token-act-pair · gpt-3.5-turbo

negations or phrases indicating the absence of something

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

former

-0.74

çļ

-0.69

Redditor

-0.69

umbn

-0.69

ourses

-0.69

send

-0.67

papers

-0.65

rift

-0.64

quet

-0.63

aviour

-0.63

POSITIVE LOGITS

 uncommon

1.37

icable

1.11

 clear

1.09

 unreasonable

1.09

 surprising

1.08

 necessarily

1.05

 easy

1.03

 advisable

0.99

 feasible

0.96

 unusual

0.95

Activations Density 0.079%

statements followed by a negation

negations or phrases indicating the absence of something

No Comments

No Known Activations