INDEX

Explanations

words related to defense or protection

oai_token-act-pair · gpt-3.5-turbo

instances of the word "defend" or its variations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 itch

-0.71

 Plug

-0.68

 ears

-0.63

 pulse

-0.62

Add

-0.62

 Measure

-0.61

 headsets

-0.61

 notice

-0.61

 scale

-0.60

goo

-0.60

POSITIVE LOGITS

 defended

3.60

 defends

2.51

 defend

2.03

 defending

1.89

 countered

1.52

 contested

1.49

 resisted

1.44

 conceded

1.44

 fought

1.44

 upheld

1.42

Activations Density 0.009%

words related to defense or protection

instances of the word "defend" or its variations

No Comments

No Known Activations