INDEX

Explanations

words related to accidents or incidents

oai_token-act-pair · gpt-3.5-turbo

phrases indicating events that occur after a specific action or incident

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

aez

-0.76

âĶĢâĶĢâĶĢâĶĢâĶĢâĶĢâĶĢâĶĢ

-0.74

achel

-0.72

tick

-0.70

NRS

-0.70

âĸĪâĸĪ

-0.68

Mask

-0.68

cci

-0.67

OO

-0.67

odiac

-0.65

POSITIVE LOGITS

noon

1.00

market

0.95

 completing

0.91

 finishing

0.87

 inspecting

0.85

 acquiring

0.83

 midnight

0.82

 receiving

0.82

 spotting

0.82

 installing

0.82

Activations Density 0.113%

words related to accidents or incidents

phrases indicating events that occur after a specific action or incident

No Comments

No Known Activations