INDEX

Explanations

organizations, departments, and specific titles within a governmental context

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 onstage

-0.70

 Sina

-0.68

 trilogy

-0.68

obin

-0.67

 epoch

-0.67

 unborn

-0.66

 capsule

-0.66

nikov

-0.66

 suffix

-0.66

 meme

-0.64

POSITIVE LOGITS

 Commissioner

1.32

 inspector

1.23

 Enforcement

1.23

 Inspect

1.23

 commissioner

1.20

 officials

1.19

 Inspector

1.18

 enforcement

1.17

 inspectors

1.17

 commissioners

1.16

Activations Density 0.342%

organizations, departments, and specific titles within a governmental context

No Comments

No Known Activations