INDEX

Explanations

phrases related to doing actions or tasks

oai_token-act-pair · gpt-3.5-turbo

occurrences of legal or ethical issues and interactions among individuals

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

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

WARN

-0.63

BuyableInstoreAndOnline

-0.61

DEBUG

-0.55

Sund

-0.55

 referen

-0.51

laun

-0.50

oret

-0.50

Osw

-0.50

çīĪ

-0.50

 Hawks

-0.50

POSITIVE LOGITS

etc

0.96

0.86

.;

0.84

!.

0.81

 respectively

0.79

.</

0.79

 whilst

0.78

.''.

0.78

.",

0.78

ãĢĤ

0.77

Activations Density 0.907%

phrases related to doing actions or tasks

occurrences of legal or ethical issues and interactions among individuals

No Comments

No Known Activations