
Translated 发布 Lara Mini 模型,活跃参数最高 2B,在 26 项基准上超越 36 个模型
来源标题:Lara Mini tops 36 models with up to 2B active params on 26 benchmarks
翻译公司 Translated 发布 Lara Mini 系列模型,在 26 项基准测试中表现优于 36 个对比模型,活跃参数最高达 2B。该模型基于混合专家与门控 Delta 网络架构,支持 44 种语言和长上下文,权重以 translated/Lara-4-Mini-Base 名义发布,采用 cc-by-nc-sa-4.0 许可。
以下正文同步自 Hacker News 最新,版权归原站所有,已转换为易读排版。
Like 11
Follow
Translated 22
Text Generation
Transformers
Safetensors
20 datasets
44 languages
lara_mini
base-model
mixture-of-experts
hybrid
gated-delta-net
multilingual
long-context
custom_code
License: cc-by-nc-sa-4.0
Model card Files Files and versions xet
Community
Deploy
Copy to bucket
Use this model Instructions to use translated/Lara-4-Mini-Base with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
Libraries
Transformers How to use translated/Lara-4-Mini-Base with Transformers:
Use a pipeline as a high-level helper
from transformers import pipeline
pipe = pipeline("text-generation", model="translated/Lara-4-Mini-Base", trust_remote_code=True)
pip install -U transformers accelerate
Load model directly
from transformers import AutoModelForCausalLM model = AutoModelForCausalLM.from_pretrained("translated/Lara-4-Mini-Base", trust_remote_code=True, device_map="auto")
Notebooks
Google Colab
Kaggle
Local Apps Settings
vLLM How to use translated/Lara-4-Mini-Base with vLLM:
Install vLLM from pip:
pip install vllm
Start the vLLM server:
vllm serve "translated/Lara-4-Mini-Base"
Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions"
-H "Content-Type: application/json"
--data '{
"model": "translated/Lara-4-Mini-Base",
"prompt": "Once upon a time,",
"max_tokens": 512,
"temperature": 0.5
}'
Use Docker
docker model run hf.co/translated/Lara-4-Mini-Base
SGLang How to use translated/Lara-4-Mini-Base with SGLang:
Install SGLang from pip:
pip install sglang
Start the SGLang server:
python3 -m sglang.launch_server
--model-path "translated/Lara-4-Mini-Base"
--host 0.0.0.0
--port 30000
Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions"
-H "Content-Type: application/json"
--data '{
"model": "translated/Lara-4-Mini-Base",
"prompt": "Once upon a time,",
"max_tokens": 512,
"temperature": 0.5
}'
Use Docker images
docker run --gpus all
--shm-size 32g
-p 30000:30000
-v ~/.cache/huggingface:/root/.cache/huggingface
--env "HF_TOKEN=<secret>"
--ipc=host
lmsysorg/sglang:latest
python3 -m sglang.launch_server
--model-path "translated/Lara-4-Mini-Base"
--host 0.0.0.0
--port 30000
Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions"
-H "Content-Type: application/json"
--data '{
"model": "translated/Lara-4-Mini-Base",
"prompt": "Once upon a time,",
"max_tokens": 512,
"temperature": 0.5
}'
Docker Model Runner How to use translated/Lara-4-Mini-Base with Docker Model Runner:
docker model run hf.co/translated/Lara-4-Mini-Base
Lara Mini base is released under the CC BY-NC-SA 4.0 license, for non-commercial use only. Access is granted for non-commercial purposes, such as research and reproducibility studies. Please share your name and affiliation: each request is reviewed by our team.</p>\n","classNames":"hf-sanitized hf-sanitized-T9NdGN4lz1Rw5L3xlfsxE"},"gated":"manual","profileSharing":{"username":true,"email":true},"isLoggedIn":false,"repoId":"translated/Lara-4-Mini-Base","repoType":"model","isGoogleGemma":false,"requiresPaidPlan":false}"> You need to agree to share your Hugging Face information to access this model
This repository is publicly accessible, but you have to accept the conditions to access its files and content.
Lara Mini base is released under the CC BY-NC-SA 4.0 license, for non-commercial use only. Access is granted for non-commercial purposes, such as research and reproducibility studies. Please share your name and affiliation: each request is reviewed by our team.
Lara Mini base is a multilingual mixture-of-experts base model by Translated, with 7.2B total and 0.6B active parameters per token, pretrained from scratch on 6T tokens and extended to a context of 128k tokens.
Overall and domain averages of Lara Mini base and of the four best other models with up to 2B active parameters on each average.
This is a base model: it continues text. It is not tuned to follow instructions or to chat, and it has no safety tuning.
Languages. English, Arabic, Azerbaijani, Bengali, Bosnian, Bulgarian, Catalan, Chinese (Simplified and Traditional), Croatian, Czech, Danish, Dutch, Estonian, Finnish, French, German, Greek, Hebrew, Hindi, Hungarian, Indonesian, Irish, Italian, Japanese, Korean, Latvian, Lithuanian, Malay, Maltese, Norwegian, Persian, Polish, Portuguese, Romanian, Russian, Serbian, Slovak, Slovenian, Spanish, Swedish, Thai, Turkish, Ukrainian and Vietnamese.
Training
Data. All training data come from publicly available datasets (some of them released under access agreements), which we deduplicated and filtered into a pool of 13.8T tokens. Training read 6T tokens from this pool, 97% of them unique. The anneal stage skips the training sequences that contain documents matching items of our evaluation benchmarks.
We release no training data with this model.
Stages. Pretraining has three stages (main, transition and anneal) over 6T tokens at a sequence length of 4,096. Each stage moves toward higher-quality, question-answer, code and math data, while multilingual text stays at about a third. A fourth stage of 100B tokens, in sequences of 32,768 and then 131,072 tokens, extends the context to 128k.
Recipe. Muon for the weight matrices and AdamW for the other parameters, peak learning rate 8e-4, global batch of about 8.4M tokens, constant learning rate for 90% of training and then a decay to zero.
Evaluation
We evaluated Lara Mini base and 56 open base models ourselves, all with the same prompts, settings and answer extraction (lm-evaluation-harness, few-shot or zero-shot, no chat template).
Overall is the mean of the four domain averages. Models are grouped by active parameters.
Tokens: pretraining tokens reported by each developer (--: not reported; †: distilled from a larger model during pretraining, wholly or in part). For Lara Mini base, the 6T of the first three stages; the long-context stage adds 100B. Gemma-4 models are listed at their effective sizes.
Speed. We measured the generation speed of 45 of these models on one A100 GPU, in bf16 without quantization. Speed is in characters per second, so that models with tokenizers of different density compare fairly. Every model with a higher Overall average than Lara Mini base generates text at least 1.8 times more slowly.
Overall average against generation speed for a single user (characters per second, log scale) for Lara Mini base and 45 open base models; the line is the Pareto frontier.
Long context. RULER (13 English tasks), ONERULER (4 retrieval tasks in 26 languages) and HELMET (4 categories), with tasks whose score does not depend on the text a base model generates after its answer.
Usage
import torch from transformers import AutoModelForCausalLM, AutoTokenizer
path = "translated/Lara-4-Mini-Base" tokenizer = AutoTokenizer.from_pretrained(path, trust_remote_code=True) model = AutoModelForCausalLM.from_pretrained(path, trust_remote_code=True, dtype=torch.bfloat16, device_map="cuda")
inputs = tokenizer("La Divina Commedia è un poema scritto da", return_tensors="pt").to(model.device) out = model.generate(**inputs, max_new_tokens=32, do_sample=False) print(tokenizer.decode(out[0], skip_special_tokens=True))
Requires transformers>=5.14 (tested up to 5.22.2).
Install flash-linear-attention for fast Gated DeltaNet kernels on GPU. Without it, a plain PyTorch path is used, which is exact but slow on long inputs.
The model was trained without a BOS token: do not prepend one. The tokenizer already does not add it.
To fine-tune, load with output_router_logits=True. The loss then includes the router load-balancing loss (router_aux_loss_coef, 0.001 as in pretraining), which keeps the experts balanced.
Limitations
A base model: it continues text. It does not follow instructions or hold a conversation, and it has no safety tuning. It can produce text that is wrong, biased or offensive.
Behavior beyond 131,072 tokens is untested.
Third-party terms that may apply
Part of the training data is synthetic text that NVIDIA generated with third-party models. The cards of these datasets state that a model trained on them "may be subject to redistribution and use requirements" in the licenses of the models that generated the data:
License
Lara Mini base is released under the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International license (CC BY-NC-SA 4.0); see LICENSE. You may share and adapt it for non-commercial purposes, giving appropriate credit, and must distribute adaptations under the same license. The code and the tokenizer are under the Apache License 2.0 (see NOTICE and LICENSE-APACHE-2.0).
For commercial use or other uses not covered by this license, we are open to licensing Lara Mini base under more permissive terms. If you are interested, please contact us at research@translated.com.
Citation
@techreport{translated2026laraminibase, title = {Lara Mini Base}, author = {{Translated, Lara Team}}, institution = {Translated}, year = {2026}, month = nov, type = {Technical Report}, url = {https://huggingface.co/translated/Lara-4-Mini-Base} }
Downloads last month -
","unk_token":null,"pad_token":"<|endoftext|>"}},"createdAt":"2026-11-11T14:38:32.000Z","discussionsDisabled":false,"discussionsSorting":"recently-created","downloads":0,"downloadsAllTime":0,"id":"translated/Lara-4-Mini-Base","isLikedByUser":false,"availableInferenceProviders":[],"showHuggingChatEntry":false,"inference":"","lastModified":"2026-11-12T14:59:58.000Z","likes":11,"pipeline_tag":"text-generation","library_name":"transformers","librariesOther":[],"trackDownloads":true,"model-index":null,"private":false,"repoType":"model","gated":"manual","gatedProfileSharing":{"username":true,"email":true},"tags":["transformers","safetensors","lara_mini","text-generation","base-model","mixture-of-experts","hybrid","gated-delta-net","multilingual","long-context","custom_code","en","ar","az","bg","bn","bs","ca","cs","da","de","el","es","et","fa","fi","fr","ga","he","hi","hr","hu","id","it","ja","ko","lt","lv","ms","mt","nl","no","pl","pt","ro","ru","sk","sl","sr","sv","th","tr","uk","vi","zh","dataset:HuggingFaceFW/fineweb-edu","dataset:HuggingFaceFW/fineweb-2","dataset:epfml/FineWeb2-HQ","dataset:epfml/FineWeb2-HQ-QA","dataset:HuggingFaceFW/finepdfs","dataset:HuggingFaceFW/finewiki","dataset:HPLT/HPLT3.0","dataset:uonlp/CulturaX","dataset:HuggingFaceCode/stack-v3-train","dataset:HuggingFaceTB/finemath","dataset:IFM/MegaMath","dataset:nvidia/Nemotron-CC-v2","dataset:nvidia/Nemotron-CC-Math-v1","dataset:nvidia/Nemotron-CC-Code-v1","dataset:nvidia/Nemotron-Pretraining-Code-v2","dataset:nvidia/Nemotron-Pretraining-SFT-v1","dataset:nvidia/Nemotron-Pretraining-Specialized-v1","dataset:nvidia/Nemotron-Pretraining-Specialized-v1.1","dataset:nvidia/Nemotron-Pretraining-Specialized-v1.2","dataset:nvidia/Nemotron-Pretraining-Legal-v1","license:cc-by-nc-sa-4.0","region:us"],"tag_objs":[{"id":"text-generation","label":"Text Generation","type":"pipeline_tag","subType":"nlp"},{"id":"transformers","label":"Transformers","type":"library"},{"id":"safetensors","label":"Safetensors","type":"library"},{"id":"dataset:HuggingFaceFW/fineweb-edu","label":"HuggingFaceFW/fineweb-edu","type":"dataset","extra":{"disabled":false}},{"id":"dataset:HuggingFaceFW/fineweb-2","label":"HuggingFaceFW/fineweb-2","type":"dataset","extra":{"disabled":false}},{"id":"dataset:epfml/FineWeb2-HQ","label":"epfml/FineWeb2-HQ","type":"dataset","extra":{"disabled":false}},{"id":"dataset:epfml/FineWeb2-HQ-QA","label":"epfml/FineWeb2-HQ-QA","type":"dataset","extra":{"disabled":false}},{"id":"dataset:Hugging
正文由 探How 从来源站点 RSS 同步,内容未经改写;遇到排版缺失或需要图片、视频时请以原文为准。