INDEX

Explanations

references related to sexual misconduct or assault

oai_token-act-pair · gpt-3.5-turbo

terms related to sexual misconduct or explicit content

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

quickShipAvailable

-0.84

 GOODMAN

-0.84

Dispatch

-0.73

 respir

-0.70

arta

-0.66

 Krug

-0.66

IELD

-0.65

 Geneva

-0.65

 Blizz

-0.63

 Solitaire

-0.63

POSITIVE LOGITS

 transmitted

0.97

 assaulted

0.88

 assaulting

0.85

 abusing

0.80

Sexual

0.80

icious

0.79

 oriented

0.79

 ensl

0.76

 attracted

0.74

mol

0.74

Activations Density 0.007%

references related to sexual misconduct or assault

terms related to sexual misconduct or explicit content

No Comments

No Known Activations