INDEX

Explanations

phrases related to characters or names from various fictional stories

oai_token-act-pair · gpt-3.5-turbo

character names in a narrative context

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

ancial

-0.84

olutions

-0.79

iners

-0.77

atories

-0.75

Companies

-0.75

Questions

-0.74

estyles

-0.73

oms

-0.73

politics

-0.73

odes

-0.72

POSITIVE LOGITS

 whom

1.23

who

1.22

 whose

1.10

who

1.06

aka

0.98

whose

0.96

 clad

0.95

 dressed

0.87

 Duchess

0.86

 blinded

0.86

Activations Density 0.226%

phrases related to characters or names from various fictional stories

character names in a narrative context

No Comments

No Known Activations