INDEX

Explanations

names, especially ones that are repeated multiple times

oai_token-act-pair · gpt-3.5-turbo

proper nouns, particularly names

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ivals

-0.86

atchewan

-0.81

ainment

-0.75

orters

-0.74

[+

-0.73

ãĥķãĤ©

-0.72

urgical

-0.71

ItemTracker

-0.71

urgy

-0.71

ablishment

-0.68

POSITIVE LOGITS

Dee

1.41

zie

0.88

ples

0.85

 Reeves

0.83

leigh

0.80

pling

0.80

pee

0.76

 Dodd

0.72

ffe

0.72

ble

0.72

Activations Density 0.007%

names, especially ones that are repeated multiple times

proper nouns, particularly names

No Comments

No Known Activations