INDEX

Explanations

phrases related to historical events or statements

oai_token-act-pair · gpt-3.5-turbo

concepts related to history, fate, and luck

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

CI

-0.74

tein

-0.71

cart

-0.64

 subsistence

-0.63

 pseudonym

-0.63

TAG

-0.61

achable

-0.61

assembly

-0.61

contact

-0.60

SHIP

-0.59

POSITIVE LOGITS

 dictates

1.37

 intervened

1.11

 prevailed

1.11

 interven

1.08

 tells

1.04

 dictate

1.00

 wills

0.98

 bends

0.96

 dictated

0.95

 informs

0.95

Activations Density 0.209%

phrases related to historical events or statements

concepts related to history, fate, and luck

No Comments

No Known Activations