INDEX

Explanations

commands or calls to action within text

oai_token-act-pair · gpt-3.5-turbo

references to the act of reading

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

ortality

-0.80

VC

-0.77

ascal

-0.71

xon

-0.69

00200000

-0.69

VP

-0.68

pload

-0.68

ugal

-0.66

unity

-0.66

ega

-0.65

POSITIVE LOGITS

 comprehension

1.13

 aloud

1.09

 reading

0.91

 readings

0.87

ied

0.83

 Reading

0.83

just

0.80

reading

0.80

itatively

0.74

reads

0.74

Activations Density 0.015%

commands or calls to action within text

references to the act of reading

No Comments

No Known Activations