INDEX

Explanations

words related to technology and coding, particularly focusing on error messages and programming terms

oai_token-act-pair · gpt-3.5-turbo

active verbs describing actions related to attacks or interventions

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

 Canaver

-0.69

ãĤ¦ãĤ¹

-0.69

ãĤ´ãĥ³

-0.68

okÃ©

-0.64

 organisers

-0.55

podcast

-0.55

 âĵĺ

-0.54

iquette

-0.53

ecided

-0.53

inburgh

-0.53

POSITIVE LOGITS

)."

0.63

)</

0.55

");

0.48

.""

0.48

 âĪ¼

0.47

 subsidized

0.46

 satur

0.45

|--

0.45

 nonexistent

0.45

 ..............

0.44

Activations Density 1.246%

words related to technology and coding, particularly focusing on error messages and programming terms

active verbs describing actions related to attacks or interventions

No Comments

No Known Activations