INDEX

Explanations

sequences of hyphens meant to indicate intensification or emphasis

oai_token-act-pair · gpt-3.5-turbo

phrases that involve time-related references

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

 Dickinson

-0.77

 Rounds

-0.70

 Bryan

-0.66

 marked

-0.65

wcs

-0.64

 Kenny

-0.64

 Practice

-0.63

¿½

-0.63

 Reborn

-0.62

Ń·

-0.61

POSITIVE LOGITS

det

0.96

themed

0.96

based

0.95

sized

0.95

cond

0.95

level

0.95

cell

0.94

oriented

0.94

transfer

0.92

chan

0.91

Activations Density 0.240%

sequences of hyphens meant to indicate intensification or emphasis

phrases that involve time-related references

No Comments

No Known Activations