INDEX

Explanations

names of users/posts and the word "posts"

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "posts."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

isin

-0.75

 Coch

-0.69

STON

-0.69

 Brotherhood

-0.68

 Orchestra

-0.68

 Heritage

-0.68

 Palest

-0.66

 Brill

-0.66

 Quint

-0.65

Kid

-0.65

POSITIVE LOGITS

 posts

1.32

posts

1.11

 postings

1.10

gres

1.05

mortem

1.04

 Posts

0.95

posted

0.94

doctoral

0.90

post

0.88

Posts

0.87

Activations Density 0.005%

names of users/posts and the word "posts"

occurrences of the word "posts."

No Comments

No Known Activations