INDEX

Explanations

details related to incidents or events that have occurred

oai_token-act-pair · gpt-3.5-turbo

instances of the word "after" indicating a subsequent event or action

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

aez

-0.81

Mask

-0.76

tick

-0.75

achel

-0.71

Near

-0.71

NRS

-0.70

âĶĢâĶĢâĶĢâĶĢâĶĢâĶĢâĶĢâĶĢ

-0.70

OO

-0.67

voy

-0.66

odiac

-0.66

POSITIVE LOGITS

 completing

0.91

market

0.89

noon

0.88

 spotting

0.84

 discovering

0.83

 acquiring

0.82

 finishing

0.82

 midnight

0.81

 receiving

0.81

 inspecting

0.81

Activations Density 0.110%

details related to incidents or events that have occurred

instances of the word "after" indicating a subsequent event or action

No Comments

No Known Activations