Benhao Tang PRO

benhaotang

AI & ML interests

Physics Master student in theoretical particle physics at Universität Heidelberg, actively looking into the possibilities of integrating AI into future physics research.

Recent Activity

posted an update 29 minutes ago

I forked and modified a version of Open DeepResearcher that other than web parsing can be done completely locally, based on Matt Schumer's work (yes, I know he is the reflection-70B guy, but this is still simple and cool) check it out at https://github.com/benhaotang/OpenDeepResearcher-via-searxng/tree/main

upvoted a collection about 13 hours ago

Paper-to-Read

upvoted an article 1 day ago

Open-R1: Update #1

View all activity

Organizations

None yet

benhaotang's activity

posted an update 29 minutes ago

Post

I forked and modified a version of Open DeepResearcher that other than web parsing can be done completely locally, based on Matt Schumer's work (yes, I know he is the reflection-70B guy, but this is still simple and cool)
check it out at
https://github.com/benhaotang/OpenDeepResearcher-via-searxng/tree/main

upvoted a collection about 13 hours ago

Paper-to-Read

Collection

6 items • Updated 1 day ago • 3

upvoted an article 1 day ago

Article

Open-R1: Update #1

•

3 days ago

• 212

upvoted a paper 1 day ago

Can We Generate Images with CoT? Let's Verify and Reinforce Image Generation Step by Step

Paper • 2501.13926 • Published 12 days ago • 33

upvoted an article 7 days ago

Article

🐺🐦‍⬛ LLM Comparison/Test: Phi-4, Qwen2 VL 72B Instruct, Aya Expanse 32B in my updated MMLU-Pro CS benchmark

•

25 days ago

• 4

liked 2 models 7 days ago

arcee-ai/Virtuoso-Lite

Text Generation • Updated 1 day ago • 512 • 26

mradermacher/Virtuoso-Lite-GGUF

Updated 4 days ago • 737 • 2

New activity in huggingface/HuggingDiscussions 7 days ago

[FEEDBACK] Inference Providers

#49 opened 18 days ago by

julien-c

reacted to mitkox's post with 👍 11 days ago

Post

2133

llama.cpp is 26.8% faster than ollama.
I have upgraded both, and using the same settings, I am running the same DeepSeek R1 Distill 1.5B on the same hardware. It's an Apples to Apples comparison.

Total duration:
llama.cpp 6.85 sec <- 26.8% faster
ollama 8.69 sec

Breakdown by phase:
Model loading
llama.cpp 241 ms <- 2x faster
ollama 553 ms

Prompt processing
llama.cpp 416.04 tokens/s with an eval time 45.67 ms <- 10x faster
ollama 42.17 tokens/s with an eval time of 498 ms

Token generation
llama.cpp 137.79 tokens/s with an eval time 6.62 sec <- 13% faster
ollama 122.07 tokens/s with an eval time 7.64 sec

llama.cpp is LLM inference in C/C++; ollama adds abstraction layers and marketing.

Make sure you own your AI. AI in the cloud is not aligned with you; it's aligned with the company that owns it.

7 replies

reacted to sometimesanotion's post with 🔥 12 days ago

Post

2699

I've managed a #1 score of 41.22% average for 14B parameter models on the Open LLM Leaderboard. As of this writing, sometimesanotion/Lamarck-14B-v0.7 is #8 for all models up to 70B parameters.

It took a custom toolchain around Arcee AI's mergekit to manage the complex merges, gradients, and LoRAs required to make this happen. I really like seeing features of many quality finetunes in one solid generalist model.