INDEX

Explanations

exact phrases or exact sequences of words

oai_token-act-pair · gpt-3.5-turbo

references to governmental actions and oversight

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 whilst

-1.15

iour

-0.94

 organise

-0.92

thood

-0.90

 Whilst

-0.89

Ò

-0.86

 realise

-0.82

avour

-0.82

 Firstly

-0.81

morrow

-0.81

POSITIVE LOGITS

 feds

1.42

DOJ

1.41

 Pentagon

1.40

 Obama

1.36

 Congressional

1.36

FBI

1.34

 ACLU

1.33

 Department

1.31

IRS

1.27

 federal

1.26

Activations Density 0.475%

exact phrases or exact sequences of words

references to governmental actions and oversight

No Comments

No Known Activations