INDEX

Explanations

words prefixed with 'ex-' or ‘ex’ at various activation levels

oai_token-act-pair · gpt-3.5-turbo

instances of the prefix "ex-" to indicate various contexts of things or people related to a former status

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 Sabha

-0.78

arity

-0.76

veyard

-0.76

 GOODMAN

-0.75

 Fever

-0.74

 Grounds

-0.69

ãĤ¤ãĥĪ

-0.69

gaard

-0.69

geist

-0.67

ysc

-0.66

POSITIVE LOGITS

ogenous

1.05

uber

0.99

clud

0.98

clamation

0.88

orbit

0.85

cerpt

0.84

oplan

0.83

terior

0.83

uding

0.81

ex

0.79

Activations Density 0.006%

words prefixed with 'ex-' or ‘ex’ at various activation levels

instances of the prefix "ex-" to indicate various contexts of things or people related to a former status

No Comments

No Known Activations