INDEX

Explanations

mentions of political figures or positions of authority

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "prime."

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

sie

-0.80

UI

-0.80

iasis

-0.70

hoe

-0.65

Alert

-0.62

 Died

-0.62

Dispatch

-0.61

 Courtesy

-0.61

alties

-0.61

 Injury

-0.61

POSITIVE LOGITS

 prime

3.84

prime

2.69

 Prime

2.37

Prime

2.35

 premier

1.61

PM

1.35

 prim

1.31

 cardinal

1.17

 king

1.01

 peak

0.99

Activations Density 0.005%

mentions of political figures or positions of authority

occurrences of the word "prime."

No Comments

No Known Activations