INDEX

Explanations

information about events or activities being held

oai_token-act-pair · gpt-3.5-turbo

instances of the word "held" in various contexts

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

————

-0.72

stasy

-0.64

clerosis

-0.63

 Conquer

-0.61

 adventurer

-0.59

zyk

-0.58

 Actress

-0.58

 deceive

-0.58

 Journey

-0.58

ettel

-0.57

POSITIVE LOGITS

 accountable

1.16

 hostage

1.15

 captive

1.01

erness

0.78

DOWN

0.77

ĸļ

0.74

overs

0.73

hold

0.73

holding

0.73

 sway

0.73

Activations Density 0.030%

information about events or activities being held

instances of the word "held" in various contexts

No Comments

No Known Activations