INDEX

Explanations

negative sentiments or problems in context, possibly related to politics or investigations

oai_token-act-pair · gpt-3.5-turbo

negations or words indicating absence

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

 Confederation

-0.71

 Observatory

-0.61

 welcome

-0.58

 Democracy

-0.58

 provisional

-0.57

 yields

-0.57

 comfort

-0.57

 Receiver

-0.57

 Conduct

-0.55

 otherwise

-0.55

POSITIVE LOGITS

Ton

0.86

raq

0.83

gel

0.81

Tan

0.80

STAR

0.80

Pak

0.80

Am

0.79

Bon

0.79

gil

0.78

Nut

0.78

Activations Density 0.084%

negative sentiments or problems in context, possibly related to politics or investigations

negations or words indicating absence

No Comments

No Known Activations