INDEX

Explanations

specific names or terms within legal, investigative, or scandalous contexts

oai_token-act-pair · gpt-3.5-turbo

references to legal cases or investigations

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

arger

-0.81

POSE

-0.71

York

-0.67

yrics

-0.66

raved

-0.63

fitting

-0.62

Higher

-0.60

ordinate

-0.60

ricular

-0.60

angible

-0.60

POSITIVE LOGITS

 debacle

0.91

 fiasco

0.84

undrum

0.80

ilogy

0.78

 spree

0.71

 saga

0.71

 trilogy

0.69

 bandwagon

0.68

 blah

0.67

 Quote

0.64

Activations Density 0.399%

specific names or terms within legal, investigative, or scandalous contexts

references to legal cases or investigations

No Comments

No Known Activations