INDEX

Explanations

events or incidents involving casualties or fatalities

oai_token-act-pair · gpt-3.5-turbo

instances of the word "that" in various contexts related to events or conditions

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

Eye

-0.64

 Corps

-0.62

 Doctrine

-0.62

ãĤĬ

-0.62

ono

-0.60

WARE

-0.59

 guessing

-0.59

 Wife

-0.56

 Physical

-0.56

ji

-0.56

POSITIVE LOGITS

 resulted

1.03

 culminated

0.97

 arose

0.89

 sparked

0.87

 amounted

0.87

 preceded

0.86

amar

0.85

 accompanies

0.83

 exceeds

0.81

 culmin

0.81

Activations Density 0.270%

events or incidents involving casualties or fatalities

instances of the word "that" in various contexts related to events or conditions

No Comments

No Known Activations