INDEX

Explanations

software-related terms and technology references

oai_token-act-pair · gpt-3.5-turbo

references to organizations or technologies

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

âĢİ

-0.72

 ..........

-0.63

lihood

-0.62

 âĢº

-0.61

hement

-0.60

javascript

-0.60

hap

-0.59

hots

-0.58

aci

-0.56

cv

-0.55

POSITIVE LOGITS

 oneself

0.65

 herself

0.65

 immortality

0.65

 himself

0.61

 identifiers

0.60

topia

0.58

 Citadel

0.57

 themselves

0.56

 detractors

0.56

 uniqueness

0.56

Activations Density 0.814%

software-related terms and technology references

references to organizations or technologies

No Comments

No Known Activations