INDEX

Explanations

references to a specific brand or product name

oai_token-act-pair · gpt-3.5-turbo

references to the word "Pro" and its variations in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

ãĤ¼ãĤ¦ãĤ¹

-0.79

 halls

-0.77

gow

-0.72

 Titanic

-0.70

 Cornell

-0.70

 straw

-0.69

ablishment

-0.69

 needles

-0.68

 spears

-0.67

owship

-0.66

POSITIVE LOGITS

digy

1.41

posal

1.20

secution

1.20

verbs

1.11

gression

1.10

poses

1.08

portion

1.07

blems

1.03

pose

1.02

secut

1.01

Activations Density 0.014%

references to a specific brand or product name

references to the word "Pro" and its variations in various contexts

No Comments

No Known Activations