INDEX

Explanations

terms related to commands or instructions beginning with "from."

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "from" in various contexts

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

faced

-0.77

ascript

-0.76

ocene

-0.75

merce

-0.73

ratulations

-0.72

atern

-0.72

lied

-0.72

isphere

-0.71

adata

-0.71

ierrez

-0.70

POSITIVE LOGITS

 afar

0.97

 whence

0.81

Wow

0.76

 thence

0.72

 ASUS

0.67

 Flavoring

0.67

 scratch

0.66

 Adinida

0.65

 inside

0.64

 Above

0.64

Activations Density 0.035%

terms related to commands or instructions beginning with "from."

occurrences of the word "from" in various contexts

No Comments

No Known Activations