INDEX

Explanations

letters and numbers combined together that represent technical specifications or codes

oai_token-act-pair · gpt-3.5-turbo

specific codes or identifiers related to various hardware or equipment types

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

ãĤ©

-0.69

crim

-0.68

 +---

-0.67

aneous

-0.67

--------------------------------------------------------

-0.67

iquette

-0.64

TPPStreamerBot

-0.63

Ø³

-0.62

âĸ¬

-0.60

wikipedia

-0.59

POSITIVE LOGITS

bps

0.74

imore

0.67

HS

0.67

 Lith

0.64

FG

0.63

dq

0.62

Sed

0.62

 mutant

0.62

endment

0.61

xus

0.61

Activations Density 0.062%

letters and numbers combined together that represent technical specifications or codes

specific codes or identifiers related to various hardware or equipment types

No Comments

No Known Activations