INDEX

Explanations

names, primarily "Jamie Dimon" and "JPMorgan"

oai_token-act-pair · gpt-3.5-turbo

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

iary

-0.93

esan

-0.86

ESCO

-0.78

hetical

-0.77

yp

-0.76

itarian

-0.73

iaries

-0.72

rums

-0.72

oration

-0.72

scape

-0.72

POSITIVE LOGITS

xx

0.90

Dim

0.84

 Benn

0.80

 Oliver

0.79

 Cull

0.75

 Shepherd

0.74

 Herrera

0.72

 Isaac

0.72

 Carr

0.71

 Leigh

0.71

Activations Density 0.031%

names, primarily "Jamie Dimon" and "JPMorgan"

No Comments

No Known Activations