INDEX

Explanations

words related to explosives or incendiary devices

oai_token-act-pair · gpt-3.5-turbo

occurrences of the suffix "ov"

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

Ĥª

-0.72

dress

-0.69

 cane

-0.66

 prevailing

-0.64

 prevail

-0.61

 prep

-0.60

 pigeon

-0.57

giving

-0.55

 commod

-0.55

fix

-0.55

POSITIVE LOGITS

ski

1.42

ideo

1.14

irus

1.10

ille

1.04

itz

1.00

rolet

0.99

itch

0.97

sky

0.96

arov

0.95

olt

0.95

Activations Density 0.018%

words related to explosives or incendiary devices

occurrences of the suffix "ov"

No Comments

No Known Activations