INDEX

Explanations

phrases related to competitions or events structured as a series of matches

oai_token-act-pair · gpt-3.5-turbo

phrases related to rankings or classifications

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

ĸļ

-1.00

 guiActiveUn

-0.71

ãĥ£

-0.69

clerosis

-0.67

NetMessage

-0.65

jab

-0.63

pora

-0.63

ounty

-0.62

jay

-0.61

igham

-0.61

POSITIVE LOGITS

iev

0.74

 EVER

0.71

 outcomes

0.70

 imaginable

0.69

 Worst

0.68

 recollection

0.67

tein

0.63

 Recon

0.62

owitz

0.61

 luck

0.61

Activations Density 0.123%

phrases related to competitions or events structured as a series of matches

phrases related to rankings or classifications

No Comments

No Known Activations