INDEX

Explanations

words related to providing or allowing something or someone

oai_token-act-pair · gpt-3.5-turbo

words related to permissions and how entities are treated

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

 Nissan

-0.68

 moder

-0.65

 paraph

-0.65

 bottleneck

-0.64

 surn

-0.63

 transmission

-0.62

 headed

-0.61

 podium

-0.60

ovo

-0.60

 rapp

-0.59

POSITIVE LOGITS

allows

2.34

 treats

2.08

allow

2.05

allowed

1.25

 Hazel

1.10

Allow

1.09

 preserves

1.07

helps

1.05

makes

1.03

 Treat

1.01

Activations Density 0.011%

words related to providing or allowing something or someone

words related to permissions and how entities are treated

No Comments

No Known Activations