INDEX

Explanations

transcripts or text passages

oai_token-act-pair · gpt-3.5-turbo

references to transcripts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

ndra

-0.73

zzle

-0.67

Magikarp

-0.66

 Soda

-0.66

bledon

-0.64

yu

-0.63

 Bengal

-0.63

pton

-0.63

GPU

-0.62

ccoli

-0.62

POSITIVE LOGITS

 transcript

1.10

 transcripts

1.05

 transcription

1.03

 Transcript

0.98

ions

0.92

ophone

0.91

ographer

0.86

 snippets

0.85

 excerpts

0.83

 taped

0.79

Activations Density 0.050%

transcripts or text passages

references to transcripts

No Comments

No Known Activations