INDEX

Explanations

numerical ranges or intervals

oai_token-act-pair · gpt-3.5-turbo

adverbs or adjectives indicating closeness or approximation

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

ngth

-0.79

rieve

-0.78

INGS

-0.72

aina

-0.69

 Shad

-0.69

RTX

-0.66

neys

-0.65

engers

-0.64

 nicely

-0.64

etter

-0.64

POSITIVE LOGITS

 nonexistent

1.18

 unanim

1.16

 impossibility

1.10

 monopoly

1.08

 unanimous

1.03

 impossible

1.03

 extinction

1.01

zero

0.97

 unstoppable

0.95

 indistinguishable

0.89

Activations Density 0.159%

numerical ranges or intervals

adverbs or adjectives indicating closeness or approximation

No Comments

No Known Activations