INDEX

Explanations

words or phrases related to something hidden or underlying

oai_token-act-pair · gpt-3.5-turbo

terms related to hidden or underlying aspects

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

eln

-0.78

ãĥ£

-0.73

zai

-0.72

yah

-0.71

ebus

-0.68

itar

-0.68

DN

-0.68

Äĩ

-0.64

ãĥĹ

-0.64

atorial

-0.63

POSITIVE LOGITS

 layers

1.11

neath

1.05

coat

0.96

sea

0.94

 bridges

0.83

 coats

0.82

pants

0.81

brush

0.78

ĸļ

0.77

pins

0.77

Activations Density 0.057%

words or phrases related to something hidden or underlying

terms related to hidden or underlying aspects

No Comments

No Known Activations