Kristian Schwethelm

KristianS7

·

kschwethelm

AI & ML interests

Large Language Models

Recent Activity

authored a paper about 14 hours ago

Depth-adaptive Inference of Looped Language Models via Continuous Depth Batching

submitted a paper about 19 hours ago

Depth-adaptive Inference of Looped Language Models via Continuous Depth Batching

upvoted a paper about 19 hours ago

Depth-adaptive Inference of Looped Language Models via Continuous Depth Batching

View all activity

Organizations

New activity in KristianS7/Ouro-2.6B about 2 months ago

Transformers v5 compatibility

#1 opened about 2 months ago by

New activity in KristianS7/Ouro-1.4B 4 months ago

Add assistant generation tags to chat template

#4 opened 4 months ago by

Update tied weight metadata for Transformers 5

#3 opened 4 months ago by

Update Ouro remote code for Transformers 5.9

#2 opened 4 months ago by

Fix default RoPE initialization for Transformers 5.9

#1 opened 4 months ago by

New activity in ByteDance/Ouro-2.6B 7 months ago

Lower evaluation results

#2 opened 10 months ago by

New activity in ByteDance/Ouro-1.4B 7 months ago

Differences in the results of the reproduction test on lm-evaluation-harness

#8 opened 9 months ago by

New activity in ByteDance/Ouro-2.6B 7 months ago

Fix bos/eos token IDs (config.json + tokenizer_config.json)

#5 opened 7 months ago by

Fix UniversalTransformerCache.get_mask_sizes for batched generation

#4 opened 7 months ago by

New activity in ByteDance/Ouro-1.4B 7 months ago

Fix bos/eos token IDs (config.json + tokenizer_config.json)

#11 opened 7 months ago by

Fix UniversalTransformerCache.get_mask_sizes for batched generation

#10 opened 7 months ago by

New activity in ByteDance/Ouro-1.4B-Thinking 7 months ago

Fix UniversalTransformerCache.get_mask_sizes for batched generation

#5 opened 7 months ago by

New activity in ByteDance/Ouro-2.6B-Thinking 7 months ago

Fix UniversalTransformerCache.get_mask_sizes for batched generation

#8 opened 7 months ago by

New activity in ByteDance/Ouro-1.4B 7 months ago

Batched generation (batch_size > 1) produces incorrect outputs — possible causal mask issue?

#9 opened 8 months ago by

New activity in ByteDance/Ouro-2.6B-Thinking 7 months ago

Fix bos/eos token IDs + add enable_thinking to chat template

#7 opened 7 months ago by

New activity in ByteDance/Ouro-1.4B-Thinking 7 months ago

Fix bos/eos token IDs + add enable_thinking to chat template

#4 opened 7 months ago by

New activity in ByteDance/Ouro-2.6B-Thinking 7 months ago

Fix bos/eos token IDs + add enable_thinking to chat template

#7 opened 7 months ago by

New activity in ByteDance/Ouro-1.4B-Thinking 7 months ago

Fix bos/eos token IDs + add enable_thinking to chat template

#4 opened 7 months ago by