kashifr
Carbon-X: the best open DNA model
Carbon-A: DNA annotation model used to discover 566M new gene candidates
Tokenizers v1
Re:DeepMind's WeatherNext model achieves breakthrough forecasting cycl...
on Saturday August 08, 2026 @10:45AM (#49222329)
Check out my pytorch reproduction of the paper here for those interested: https://github.com/NVIDIA/physicsnemo/pull/1660
Carbon: Autoregressive Genomic Foundation Model
The ultimate guide to RL environments: building and scaling them in the LLM era
Distilling 100B+ Models 40x Faster with TRL
Re:TinyLoRA – Learning to Reason in 13 Parameters
on Wednesday April 01, 2026 @09:34AM (#47600653)
You can try out TinyLoRA in PEFT main now: https://huggingface.co/docs/peft/main/en/package_reference/t...
Keep the Tokens Flowing: Lessons from 16 Open-Source RL Libraries
Transformers V5 is out!
The Smol Training Playbook: The Secrets to Building World-Class LLMs
Unlocking On-Policy Distillation for Any Model Family
Transformers 4.55 New OpenAI GPT OSS
Smollm3: Smol, multilingual, long-context reasoner LLM
Re:DuckDB is probably the most important geospatial software of the la...
on Sunday May 04, 2025 @07:03AM (#43886016)
Have you tried https://geobase.app/ they recently also had a post about duckdb integration: https://geobase.app/blog/duckdb-1-1-3
Epic vs. Apple
AIMO (AI Math Olympiad) progress prize winning solution
MaPO: A reference-free alignment technique for diffusion models
Re:MaPO: A reference-free alignment technique for diffusion models
on Tuesday June 11, 2024 @04:52AM (#40644007)
A novel and memory-friendly preference alignment method for diffusion models that does not depend on any reference model.
Re:OpenHermesPreferences: Dataset of ~1M AI preferences from teknium/O...
on Monday February 26, 2024 @11:56AM (#39513674)
The dataset can be used for training preference models or aligning language models through techniques like Direct Preference Optimization (DPO).
OpenHermesPreferences: Dataset of ~1M AI preferences from teknium/OpenHermes-2.5
HuggingFace Training Cluster as a Service
HuggingFace 235M series D at a $4.5B valuation
Fine-tune Llama 2 with DPO
Re:QLoRA 4-bit finetuning of LLMs
on Wednesday May 24, 2023 @01:01PM (#36060973)
An efficient finetuning approach that reduces memory usage enough to finetune a 65B parameter model on a single 48GB GPU while preserving full 16-bit finetuning task performance!
QLoRA 4-bit finetuning of LLMs
Re:V0.15.0 Beyond Image Generation
on Wednesday April 12, 2023 @03:16PM (#35545192)
The title is confusing but refers to the Huggingface Diffusers v0.15 release which brings new pipelines for video and audio to diffusers, showing that diffusion is a great choice for all sorts of generative tasks.
StackLlama: A hands-on guide to train LlaMa with RLHF
Re:StackLlama: A hands-on guide to train LlaMa with RLHF
on Thursday April 06, 2023 @03:06PM (#35472931)
All the steps involved in training a LlaMa model to answer questions on Stack Exchange data with RLHF.