INDEX

Explanations

words related to assigning blame or responsibility

oai_token-act-pair · gpt-3.5-turbo

instances of the word "blame"

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

ires

-0.75

chan

-0.74

raised

-0.70

artney

-0.70

ouver

-0.69

irl

-0.69

opened

-0.68

forms

-0.67

iring

-0.67

quire

-0.65

POSITIVE LOGITS

 blame

1.38

 blames

0.95

 blaming

0.94

 attribut

0.88

 blamed

0.87

 forgiven

0.80

 forgiveness

0.80

 culp

0.78

 culprit

0.78

 burden

0.71

Activations Density 0.006%

words related to assigning blame or responsibility

instances of the word "blame"

No Comments

No Known Activations