INDEX

Explanations

terms associated with anti-American and anti-religious sentiments

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs49152-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

49,152

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

hower

-0.77

 srfAttach

-0.72

OST

-0.71

INS

-0.71

LESS

-0.70

iths

-0.70

osc

-0.69

ename

-0.68

Begin

-0.67

kens

-0.66

POSITIVE LOGITS

 sentiment

1.14

 sentiments

1.10

 bigotry

1.03

 prejudice

0.99

 rhetoric

0.97

 violence

0.89

 bias

0.88

 ferv

0.87

 hysteria

0.86

tir

0.82

Activations Density 0.019%

terms associated with anti-American and anti-religious sentiments

No Comments

No Known Activations