Total runs:
26.3K
Run Growth:
25.6K
Growth Rate:
97.42%
Sponsored by
ZenMux
arcee-ai / Trinity-Large-Base
Model's Last Updated: April 01 2026
text-generation
Introduction of Trinity-Large-Base
Model Details of Trinity-Large-Base
Trinity-Large-Base
Introduction
Trinity-Large-Base is a pretrained foundation model from Arcee AI's Trinity Large training run. It is a 398B-parameter sparse Mixture-of-Experts (MoE) model with approximately 13B active parameters per token. The checkpoint was captured after 17 trillion tokens of pretraining, including mid-training learning-rate anneals and context extension, but prior to any instruction tuning or reinforcement learning.
This checkpoint represents the completed pretraining phase and serves as a foundation for research and downstream fine-tuning.
More details on the training of Trinity Large are available in the technical report .
Model Variants
The Trinity Large family consists of three checkpoints from the same training run:
- Trinity-Large-Base (this release): Full 17T-token pretrained foundation model with mid-training anneals
- Trinity-Large-TrueBase : 10T-token pre-anneal checkpoint with no instruction data
- Trinity-Large-Preview : Lightly post-trained, chat-ready model undergoing active RL
Architecture
Trinity-Large-Base uses a sparse MoE configuration designed to maximize efficiency while maintaining large-scale capacity.
| Hyperparameter | Value |
|---|---|
| Total parameters | ~398B |
| Active parameters per token | ~13B |
| Experts | 256 |
| Active experts | 4 |
| Routing strategy | 4-of-256 (1.56% sparsity) |
| Dense layers | 6 |
| Pretraining context length | 8,192 |
| Context length after extention | 512k |
| Architecture | Sparse MoE (AfmoeForCausalLM) |
Benchmark Results
| Benchmark | N-shot | Metric | Score | Stderr |
|---|---|---|---|---|
| mbpp_plus | 3 | pass_at_1,none | 0.8862 | ±0.0164 |
| minerva_math500 | 4 | math_verify,none | 0.6520 | ±0.0213 |
| hellaswag_5shot | 5 | acc_norm,none | 0.9011 | ±0.0030 |
| winogrande_5shot | 5 | acc,none | 0.8082 | ±0.0111 |
| mmlu_5shot | 5 | acc,none | 0.8258 | ±0.0031 |
| mmlu_generative_5shot | 5 | exact_match,get_response | 0.8260 | ±0.0031 |
| mmlu_pro | 5 | exact_match,custom-extract | 0.6602 | ±0.0042 |
| triviaqa_5shot | 5 | exact_match,remove_whitespace | 0.8330 | ±0.0028 |
| arc_challenge_0shot | 0 | acc_norm,none | 0.6544 | ±0.0139 |
| bbh_fewshot | 3 | exact_match,remove_whitespace | 0.6570 | ±0.0051 |
| gpqa_diamond_5shot | 5 | acc_norm,none | 0.4394 | ±0.0354 |
| gsm8k_cot | 8 | exact_match,flexible-extract | 0.9136 | ±0.0077 |
Training Configuration
Pretraining
- Training tokens: 17 trillion
- Checkpoint type: Post-anneal (foundation)
- Instruction data: None
- RLHF or post-training: None
This checkpoint represents the final pretrained state after completion of the pretraining phase, including mid-training learning-rate anneals, but before instruction tuning or reinforcement learning.
Optimizers
Optimizer learning rates during WSD stable phase:
- Adam learning rate: 2e-4
- Muon learning rate: 8e-4
Muon was used to support larger critical batch sizes in a highly sparse MoE regime.
Infrastructure
- Hardware: 2,048 NVIDIA B300 GPUs
- Parallelism: HSDP + Expert Parallelism
- Compute partner: Prime Intellect
- Data partner: Datology
Intended Use
- Studying emergent behavior from large-scale pretraining
- Sparse MoE routing and load-balancing research
- Interpretability, probing, and ablation studies
- Domain-specific fine-tuning from a pretrained foundation
- Academic and industrial foundation model research
Comparison with TrueBase
Trinity-Large-Base includes an additional 7 trillion training tokens compared to Trinity-Large-TrueBase, along with mid-training learning-rate anneals. These anneals stabilize training dynamics and typically improve downstream fine-tuning performance compared to the pre-anneal checkpoint. Researchers studying raw pretraining dynamics may prefer TrueBase, while those seeking a foundation for fine-tuning may prefer this checkpoint.
Known Limitations
- Not aligned for safety, helpfulness, or conversational tone
- Requires substantial compute and expertise to fine-tune
- May exhibit raw or unstable behaviors typical of unaligned models
- No extended-context tuning beyond the 8K pretraining window
License
Trinity-Large-Base is released under the Apache License, Version 2.0.
Runs of arcee-ai Trinity-Large-Base on huggingface.co
441
Total runs
0
24-hour runs
0
3-day runs
14
7-day runs
159
30-day runs
More Information About Trinity-Large-Base huggingface.co Model
Url of Trinity-Large-Base
Trinity-Large-Base huggingface.co Url
Provider of Trinity-Large-Base huggingface.co
Other API from arcee-ai
Updated:April 09 2026
Total runs:
19.9K
Run Growth:
3.5K
Growth Rate:
17.45%
Total runs:
18.7K
Run Growth:
2.4K
Growth Rate:
12.95%
Total runs:
10.0K
Run Growth:
9.3K
Growth Rate:
92.47%
Total runs:
7.4K
Run Growth:
-348
Growth Rate:
-4.72%
Total runs:
3.1K
Run Growth:
3.0K
Growth Rate:
99.61%
Total runs:
3.0K
Run Growth:
-33
Growth Rate:
-1.12%
Total runs:
2.2K
Run Growth:
682
Growth Rate:
30.56%
Total runs:
2.1K
Run Growth:
1.1K
Growth Rate:
53.31%
Total runs:
2.1K
Run Growth:
-306
Growth Rate:
-14.50%
Total runs:
1.8K
Run Growth:
1.5K
Growth Rate:
82.87%
Total runs:
1.5K
Run Growth:
1.5K
Growth Rate:
98.63%
Total runs:
1.4K
Run Growth:
10
Growth Rate:
0.71%
Total runs:
988
Run Growth:
-55
Growth Rate:
-5.57%
Total runs:
850
Run Growth:
0
Growth Rate:
0.00%
Total runs:
840
Run Growth:
631
Growth Rate:
75.12%
Total runs:
835
Run Growth:
436
Growth Rate:
52.22%
Total runs:
803
Run Growth:
-190
Growth Rate:
-23.66%
Total runs:
785
Run Growth:
-1.3K
Growth Rate:
-164.59%
Total runs:
779
Run Growth:
567
Growth Rate:
72.79%
Total runs:
766
Run Growth:
-6.3K
Growth Rate:
-820.89%
Total runs:
758
Run Growth:
480
Growth Rate:
63.32%
Total runs:
614
Run Growth:
-4.4K
Growth Rate:
-715.15%
Total runs:
568
Run Growth:
319
Growth Rate:
56.16%
Total runs:
503
Run Growth:
382
Growth Rate:
75.94%
Total runs:
408
Run Growth:
0
Growth Rate:
0.00%
Total runs:
351
Run Growth:
351
Growth Rate:
100.00%
Total runs:
281
Run Growth:
0
Growth Rate:
0.00%
Total runs:
251
Run Growth:
166
Growth Rate:
66.14%
Total runs:
221
Run Growth:
153
Growth Rate:
69.23%
Total runs:
194
Run Growth:
-460
Growth Rate:
-237.11%
Total runs:
180
Run Growth:
-1
Growth Rate:
-0.56%
Total runs:
158
Run Growth:
1
Growth Rate:
0.63%
Total runs:
148
Run Growth:
-99
Growth Rate:
-66.89%
Total runs:
141
Run Growth:
36
Growth Rate:
25.53%
Total runs:
137
Run Growth:
116
Growth Rate:
84.67%
Total runs:
131
Run Growth:
18
Growth Rate:
13.74%
Total runs:
107
Run Growth:
4
Growth Rate:
3.74%
Total runs:
102
Run Growth:
-242
Growth Rate:
-237.25%
Total runs:
100
Run Growth:
79
Growth Rate:
79.00%
Total runs:
88
Run Growth:
-19
Growth Rate:
-21.59%
Total runs:
78
Run Growth:
-23
Growth Rate:
-29.49%
Total runs:
78
Run Growth:
-66
Growth Rate:
-84.62%
Total runs:
76
Run Growth:
72
Growth Rate:
94.74%
Total runs:
73
Run Growth:
62
Growth Rate:
84.93%
Total runs:
70
Run Growth:
-1
Growth Rate:
-1.43%
Total runs:
69
Run Growth:
51
Growth Rate:
73.91%
Total runs:
66
Run Growth:
-132
Growth Rate:
-200.00%
Total runs:
55
Run Growth:
-166
Growth Rate:
-301.82%
Total runs:
46
Run Growth:
35
Growth Rate:
76.09%
Total runs:
45
Run Growth:
35
Growth Rate:
77.78%
Total runs:
37
Run Growth:
8
Growth Rate:
21.62%
Total runs:
34
Run Growth:
27
Growth Rate:
79.41%
Total runs:
31
Run Growth:
-11
Growth Rate:
-35.48%
Total runs:
23
Run Growth:
9
Growth Rate:
39.13%
Total runs:
22
Run Growth:
11
Growth Rate:
50.00%
Total runs:
20
Run Growth:
0
Growth Rate:
0.00%
Total runs:
18
Run Growth:
-18
Growth Rate:
-100.00%
Total runs:
16
Run Growth:
10
Growth Rate:
62.50%
Total runs:
15
Run Growth:
-5
Growth Rate:
-33.33%
Total runs:
15
Run Growth:
-29
Growth Rate:
-193.33%
Total runs:
14
Run Growth:
-44
Growth Rate:
-314.29%
Total runs:
14
Run Growth:
-2
Growth Rate:
-14.29%
Total runs:
13
Run Growth:
7
Growth Rate:
53.85%
Total runs:
13
Run Growth:
9
Growth Rate:
69.23%
Total runs:
13
Run Growth:
7
Growth Rate:
53.85%
Total runs:
13
Run Growth:
9
Growth Rate:
69.23%
Total runs:
13
Run Growth:
12
Growth Rate:
92.31%
Total runs:
12
Run Growth:
5
Growth Rate:
41.67%
Total runs:
12
Run Growth:
5
Growth Rate:
41.67%
Total runs:
12
Run Growth:
6
Growth Rate:
50.00%