Hugging Face
Search models, datasets, users...
Ctrl K
Models
Datasets
Spaces
Buckets
Docs
Enterprise
Pricing
Website
Tasks
HuggingChat
Collections
Languages
Organizations
Community
Blog
Posts
Daily Papers
Hardware
Learn
Discord
Forum
GitHub
Solutions
Team & Enterprise
Hugging Face PRO
Enterprise Support
Inference Providers
Inference Endpoints
Storage Buckets
Log In
Sign Up
marsnext
/
Phobos-1
Like
0
Follow
MarsNext
1
Text Generation
PyTorch
Safetensors
English
phobos_cortical_moe
marsnext
phobos-1
conversational
code-generation
math-reasoning
system-2-reasoning
mixture-of-experts
cortical-moe
dynamic-mitosis
deepinfer-engram
multi-token-prediction
openai-compatible
edge-inference
License:
apache-2.0
Model card
Files
Files and versions
xet
Community
Copy to bucket
main
Phobos-1
684 MB
Ctrl+K
Ctrl+K
1 contributor
History:
23 commits
devMartianLabs
Update architecture diagram to clean Left-to-Right execution pipeline (generate_architecture_diagram.py)
488e3cf
verified
3 days ago
assets
Update architecture diagram to clean Left-to-Right execution pipeline (assets/phobos1_architecture_diagram.svg)
3 days ago
.gitattributes
Safe
1.67 kB
Add dark-mode architecture diagram and update README (assets/phobos1_architecture_diagram.png)
3 days ago
README.md
Safe
13.3 kB
Add dark-mode architecture diagram and update README (README.md)
3 days ago
architecture.py
Safe
25.8 kB
Release MarsNext Phobos-1 (100% 35-Dataset Base Step 33977 + 60-Book 50/50 Replay Post-Trained)
3 days ago
benchmark_10prompts_comparison.json
Safe
18.8 kB
Release MarsNext Phobos-1 (100% 35-Dataset Base Step 33977 + 60-Book 50/50 Replay Post-Trained)
3 days ago
bpe_tokenizer.py
Safe
7.19 kB
Release MarsNext Phobos-1 (100% 35-Dataset Base Step 33977 + 60-Book 50/50 Replay Post-Trained)
3 days ago
chat_cli.py
Safe
9.18 kB
Ensure direct mode masks THINK_START_ID when --think is not enabled in chat_cli.py
3 days ago
config.json
Safe
536 Bytes
Release MarsNext Phobos-1 (100% 35-Dataset Base Step 33977 + 60-Book 50/50 Replay Post-Trained)
3 days ago
deepinfer_modules.py
Safe
17.5 kB
Release MarsNext Phobos-1 (100% 35-Dataset Base Step 33977 + 60-Book 50/50 Replay Post-Trained)
3 days ago
generate_architecture_diagram.py
Safe
15.9 kB
Update architecture diagram to clean Left-to-Right execution pipeline (generate_architecture_diagram.py)
3 days ago
generate_benchmark_charts.py
Safe
11.3 kB
Add Artificial-Analysis style dark-mode benchmark card (generate_benchmark_charts.py)
3 days ago
mitosis.py
Safe
7.63 kB
Release MarsNext Phobos-1 (100% 35-Dataset Base Step 33977 + 60-Book 50/50 Replay Post-Trained)
3 days ago
model.safetensors
226 MB
xet
Release MarsNext Phobos-1 (100% 35-Dataset Base Step 33977 + 60-Book 50/50 Replay Post-Trained)
3 days ago
multi_model_head_to_head_benchmark.json
Safe
7.7 kB
Add head-to-head multi-model benchmark results and scripts (multi_model_head_to_head_benchmark.json)
3 days ago
phobos1_base_100pct_step33977_bf16.pt
Safe
226 MB
xet
Release MarsNext Phobos-1 (100% 35-Dataset Base Step 33977 + 60-Book 50/50 Replay Post-Trained)
3 days ago
phobos1_full_benchmark_results.json
Safe
26.8 kB
Add head-to-head multi-model benchmark results and scripts (phobos1_full_benchmark_results.json)
3 days ago
phobos1_post_trained_bf16.pt
Safe
226 MB
xet
Release MarsNext Phobos-1 (100% 35-Dataset Base Step 33977 + 60-Book 50/50 Replay Post-Trained)
3 days ago
phobos_bpe_32k.json
Safe
2.37 MB
Release MarsNext Phobos-1 (100% 35-Dataset Base Step 33977 + 60-Book 50/50 Replay Post-Trained)
3 days ago
run_multi_model_comparison.py
Safe
17.7 kB
Add head-to-head multi-model benchmark results and scripts (run_multi_model_comparison.py)
3 days ago
run_phobos1_benchmark_suite.py
Safe
15.4 kB
Add head-to-head multi-model benchmark results and scripts (run_phobos1_benchmark_suite.py)
3 days ago
serve_openai_api.py
Safe
9.71 kB
Ensure direct mode masks THINK_START_ID when --think is not enabled in serve_openai_api.py
3 days ago
tokenizer.json
Safe
2.37 MB
Release MarsNext Phobos-1 (100% 35-Dataset Base Step 33977 + 60-Book 50/50 Replay Post-Trained)
3 days ago
training_telemetry_33977steps.json
Safe
115 kB
Release MarsNext Phobos-1 (100% 35-Dataset Base Step 33977 + 60-Book 50/50 Replay Post-Trained)
3 days ago