INDEX

Explanations

names and terms related to various individuals or entities

oai_token-act-pair · gpt-3.5-turbo

references to specific individuals or positions of authority related to political or organizational 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

æ©

-0.83

ãĤ¨ãĥ«

-0.75

 straw

-0.72

=-=-=-=-

-0.71

çīĪ

-0.70

 Ø§ÙĦ

-0.70

 Takeru

-0.69

å£

-0.69

ãĤ¼ãĤ¦ãĤ¹

-0.68

 EVENTS

-0.66

POSITIVE LOGITS

rians

0.86

closure

0.86

omaly

0.86

ression

0.78

rius

0.76

Syn

0.75

rity

0.75

uve

0.73

ury

0.73

rian

0.73

Activations Density 0.060%

names and terms related to various individuals or entities

references to specific individuals or positions of authority related to political or organizational contexts

No Comments

No Known Activations