INDEX

Explanations

references to lectures or events where someone is speaking to an audience

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "lecture" and its variants

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

bably

-0.68

adjusted

-0.68

uilt

-0.66

dating

-0.61

 Scotia

-0.61

existing

-0.59

imposed

-0.59

fortunately

-0.58

odka

-0.58

nergy

-0.58

POSITIVE LOGITS

 Lect

1.44

 lectures

1.09

ĸļ

1.04

 lecture

1.03

 lect

0.91

ysis

0.90

inar

0.86

 seminars

0.86

ļéĨĴ

0.84

¿½

0.84

Activations Density 0.007%

references to lectures or events where someone is speaking to an audience

occurrences of the word "lecture" and its variants

No Comments

No Known Activations