INDEX

Explanations

words related to governmental roles and actions, particularly those related to ministries

oai_token-act-pair · gpt-3.5-turbo

references to the word "Min" and 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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 bipolar

-0.63

 Niet

-0.63

 Arnold

-0.62

 Strauss

-0.61

 sexually

-0.60

 seismic

-0.60

 associated

-0.60

 cove

-0.60

 fabric

-0.59

 readers

-0.59

POSITIVE LOGITS

Min

3.84

min

2.03

MIN

1.94

Min

1.84

MIN

1.75

minist

1.48

min

1.43

mins

1.23

 Minutes

1.23

Minimum

1.21

Activations Density 0.010%

words related to governmental roles and actions, particularly those related to ministries

references to the word "Min" and its variations

No Comments

No Known Activations