INDEX

Explanations

terms related to a specific topic or entity, such as "Che" in this case

oai_token-act-pair · gpt-3.5-turbo

mentions of the term "Che" in various contexts

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

CTR

-0.86

ãĥ¼ãĥĨ

-0.86

ãĥ¼ãĥĨãĤ£

-0.85

 Archdemon

-0.74

 executive

-0.68

 Icelandic

-0.65

 Instruments

-0.61

 overflow

-0.61

 JUSTICE

-0.60

 Legacy

-0.60

POSITIVE LOGITS

bones

1.15

lean

1.13

hou

1.09

chet

1.01

bone

0.98

esy

0.97

apeake

0.96

roo

0.94

anut

0.93

bag

0.93

Activations Density 0.012%

terms related to a specific topic or entity, such as "Che" in this case

mentions of the term "Che" in various contexts

No Comments

No Known Activations