INDEX

Explanations

academic research-related phrases

oai_token-act-pair · gpt-3.5-turbo

instances of the verb "was" and related variations showing past actions or states

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

addons

-0.74

ardless

-0.69

mods

-0.68

 Celt

-0.67

ukong

-0.66

fighter

-0.66

Introduced

-0.65

 Afgh

-0.64

 Chili

-0.63

Cot

-0.63

POSITIVE LOGITS

 conducted

1.51

 published

1.33

 undertaken

1.23

conduct

1.14

 retrospective

1.13

 funded

1.10

 observational

1.08

 conclusive

1.07

 publish

1.06

 incon

1.06

Activations Density 0.138%

academic research-related phrases

instances of the verb "was" and related variations showing past actions or states

No Comments

No Known Activations