INDEX

Explanations

words related to military operations or locations in the Middle East, particularly focusing on terms related to conflicts in the region

oai_token-act-pair · gpt-3.5-turbo

sequences of letters that end with "jah", "bah", "yah", or similar syllable patterns typically found in Arabic or Semitic names

oai_token-act-pair · gpt-4-turbo

terms related to specific cultural or ethnic identities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Pengu

-0.78

enhagen

-0.69

Moon

-0.67

 Takeru

-0.65

uncture

-0.63

 fixation

-0.62

Sund

-0.62

 cubes

-0.61

arium

-0.61

 humanoid

-0.61

POSITIVE LOGITS

tml

1.23

ttp

0.98

awk

0.94

iyah

0.93

jah

0.92

renheit

0.89

awks

0.89

ouston

0.88

olics

0.87

keeper

0.85

Activations Density 0.013%

words related to military operations or locations in the Middle East, particularly focusing on terms related to conflicts in the region

sequences of letters that end with "jah", "bah", "yah", or similar syllable patterns typically found in Arabic or Semitic names

terms related to specific cultural or ethnic identities

No Comments

No Known Activations