INDEX

Explanations

words related to historical events and judgments

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

instances of the word "far"

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 12-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.hook_resid_post

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.11.hook_resid_post

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

 Syndicate

-0.63

 dstg

-0.60

æ©

-0.58

 Tycoon

-0.54

 Manufact

-0.54

 Sorce

-0.52

 Killing

-0.52

 Tomato

-0.52

 Magicka

-0.51

 Amos

-0.51

POSITIVE LOGITS

fetched

0.98

 flung

0.83

 back

0.82

ce

0.78

 away

0.76

as

0.69

rowing

0.67

go

0.67

med

0.67

ces

0.66

Activations Density 0.021%

words related to historical events and judgments

instances of the word "far"

No Comments

No Known Activations