INDEX

Explanations

strong emotional or decisive actions or statements related to various situations

oai_token-act-pair · gpt-3.5-turbo

words related to actions or processes commonly associated with criminal or contentious contexts

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

Disclaimer

-0.73

 absence

-0.69

 aisle

-0.68

 identification

-0.68

 divergence

-0.67

 takeaway

-0.66

 implication

-0.66

 variation

-0.64

 objective

-0.63

 outcome

-0.62

POSITIVE LOGITS

itates

1.03

ifies

1.02

izes

1.00

ously

0.92

posed

0.90

aciously

0.88

cedes

0.88

isively

0.88

anted

0.87

igrated

0.86

Activations Density 0.362%

strong emotional or decisive actions or statements related to various situations

words related to actions or processes commonly associated with criminal or contentious contexts

No Comments

No Known Activations