INDEX

Explanations

dialogues including quotes or statements

oai_token-act-pair · gpt-3.5-turbo

occurrences of the character "ľ" in dialogue

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

 targeted

-0.88

 Polk

-0.76

 accelerated

-0.75

 marked

-0.72

 Patriarch

-0.71

 territorial

-0.71

 fragmented

-0.71

 selection

-0.70

 Monarch

-0.69

 dubbed

-0.69

POSITIVE LOGITS

yeah

1.28

sorry

1.18

hello

1.18

oops

1.13

everything

1.12

why

1.12

there

1.11

wow

1.10

YOU

1.10

thank

1.10

Activations Density 0.128%

dialogues including quotes or statements

occurrences of the character "ľ" in dialogue

No Comments

No Known Activations