INDEX

Explanations

instances where the subject will be affected or influenced by something

oai_token-act-pair · gpt-3.5-turbo

instances of the word "be" in various 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

 procedure

-0.70

 entails

-0.69

 process

-0.69

 divide

-0.68

 disperse

-0.65

 scrimmage

-0.65

 unfold

-0.64

 formulation

-0.63

 converge

-0.63

 arise

-0.63

POSITIVE LOGITS

 able

1.21

friends

1.04

league

0.95

acons

0.95

heading

0.93

 thankful

0.93

 glad

0.90

ardless

0.88

 grateful

0.88

 rewarded

0.88

Activations Density 0.238%

instances where the subject will be affected or influenced by something

instances of the word "be" in various contexts

No Comments

No Known Activations