INDEX

Explanations

dialogue markers or quoted speech

oai_token-act-pair · gpt-3.5-turbo

quoted speech or dialogue

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

 upstream

-0.87

 pole

-0.78

 downstream

-0.77

 outsourcing

-0.77

 frontline

-0.76

 characterized

-0.71

 isolated

-0.70

 valued

-0.69

 headquartered

-0.69

 targeted

-0.69

POSITIVE LOGITS

Oh

1.26

Yeah

1.16

Hey

1.16

Fuck

1.14

Hmm

1.12

Yes

1.10

Sorry

1.10

Damn

1.07

Well

1.07

Alright

1.06

Activations Density 0.109%

dialogue markers or quoted speech

quoted speech or dialogue

No Comments

No Known Activations