INDEX

Explanations

terms related to medical diagnosis or political outcomes, especially with a focus on negative consequences

oai_token-act-pair · gpt-3.5-turbo

terms related to diagnosis and losing situations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 mete

-0.67

link

-0.65

 Discord

-0.62

 commun

-0.61

Nun

-0.57

gig

-0.57

 meteor

-0.57

 charm

-0.56

Fam

-0.56

å¥³

-0.56

POSITIVE LOGITS

osing

4.13

oses

2.77

osed

2.62

oser

2.05

ose

2.03

osures

1.80

OSE

1.77

OSED

1.48

osite

1.47

osis

1.43

Activations Density 0.015%

terms related to medical diagnosis or political outcomes, especially with a focus on negative consequences

terms related to diagnosis and losing situations

No Comments

No Known Activations