INDEX

Explanations

phrases related to actions and intentions of people

oai_token-act-pair · gpt-3.5-turbo

connections between actions and consequences

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

rique

-0.89

asio

-0.77

 fronts

-0.76

rers

-0.66

hoe

-0.65

lance

-0.62

 reinstated

-0.62

hei

-0.61

onde

-0.60

 holders

-0.60

POSITIVE LOGITS

 namely

0.92

viz

0.78

excluding

0.74

 Whether

0.73

Magikarp

0.71

 Including

0.70

 except

0.67

whether

0.65

 INCLUD

0.64

BUT

0.64

Activations Density 0.760%

phrases related to actions and intentions of people

connections between actions and consequences

No Comments

No Known Activations