INDEX

Explanations

statements related to putting something or someone in a dangerous or harmful situation, particularly with a focus on the concept of 'risk'

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate a state of being endangered or in jeopardy

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

anche

-0.76

elf

-0.70

ript

-0.69

encer

-0.63

isl

-0.62

rix

-0.62

MX

-0.61

ann

-0.61

ort

-0.61

frey

-0.61

POSITIVE LOGITS

ħĭ

0.86

 endanger

0.72

 starvation

0.72

 Palestin

0.70

 Survivors

0.67

 Survive

0.67

 casualties

0.66

 horm

0.65

 brink

0.65

atives

0.64

Activations Density 0.015%

statements related to putting something or someone in a dangerous or harmful situation, particularly with a focus on the concept of 'risk'

phrases that indicate a state of being endangered or in jeopardy

No Comments

No Known Activations