Text Generation
Transformers
Safetensors
llama
symbolic-music
music-generation
musicxml
midi
remi
bar-major
text-generation-inference
Instructions to use haster/Fermata-0.2B-v1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use haster/Fermata-0.2B-v1 with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="haster/Fermata-0.2B-v1")# pip install -U transformers accelerate # Load model directly from transformers import AutoTokenizer, AutoModelForCausalLM tokenizer = AutoTokenizer.from_pretrained("haster/Fermata-0.2B-v1") model = AutoModelForCausalLM.from_pretrained("haster/Fermata-0.2B-v1", device_map="auto") - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- vLLM
How to use haster/Fermata-0.2B-v1 with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "haster/Fermata-0.2B-v1" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "haster/Fermata-0.2B-v1", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'Use Docker
docker model run hf.co/haster/Fermata-0.2B-v1
- SGLang
How to use haster/Fermata-0.2B-v1 with SGLang:
Install from pip and serve model
# Install SGLang from pip: pip install sglang # Start the SGLang server: python3 -m sglang.launch_server \ --model-path "haster/Fermata-0.2B-v1" \ --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": "haster/Fermata-0.2B-v1", "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 "haster/Fermata-0.2B-v1" \ --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": "haster/Fermata-0.2B-v1", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }' - Docker Model Runner
How to use haster/Fermata-0.2B-v1 with Docker Model Runner:
docker model run hf.co/haster/Fermata-0.2B-v1
Download run_summary.json from haster/Fermata-0.2B-v1: direct link, hf CLI and curl.
- Browser
- Download file 8.65 kB
-
https://huggingface.co/haster/Fermata-0.2B-v1/resolve/main/run_summary.json
- Command line
-
hf download hf://haster/Fermata-0.2B-v1/run_summary.json
-
curl -L -o run_summary.json https://huggingface.co/haster/Fermata-0.2B-v1/resolve/main/run_summary.json
8.65 kB
| { | |
| "run": { | |
| "git_commit": null, | |
| "config": { | |
| "path": "/home/nates/musicxml/configs/train_200m_sft_tier1_vast.yaml", | |
| "sha256": "02851fe6647c34f735ad7852f1226516da6ebd56e482d920da5de5a67643d770", | |
| "values": { | |
| "model_config": "configs/model_200m_tier1.yaml", | |
| "shard_root": "data/shards/sft_tier1", | |
| "tokenizer_root": "artifacts/tokenizer_v2_tier1", | |
| "seq_len": 2048, | |
| "init_from": "artifacts/model_200m_tier1/ckpt_latest.pt", | |
| "micro_batch": 32, | |
| "grad_accum": 8, | |
| "gradient_checkpointing": false, | |
| "fp8": true, | |
| "fp8_recipe": "tensorwise", | |
| "compile": true, | |
| "dtype": "bfloat16", | |
| "max_steps": 760, | |
| "warmup_steps": 38, | |
| "lr": 3e-05, | |
| "min_lr": 3e-06, | |
| "weight_decay": 0.1, | |
| "beta1": 0.9, | |
| "beta2": 0.95, | |
| "grad_clip": 1.0, | |
| "seed": 20260910, | |
| "shuffle_data": true, | |
| "deterministic_data_order": true, | |
| "log_interval": 20, | |
| "eval_interval": 100, | |
| "eval_max_batches": 20, | |
| "checkpoint_interval": 100, | |
| "ckpt_path": "artifacts/model_200m_sft_tier1/ckpt_best.pt", | |
| "latest_ckpt_path": "artifacts/model_200m_sft_tier1/ckpt_latest.pt", | |
| "summary_path": "artifacts/model_200m_sft_tier1/run_summary.json" | |
| } | |
| }, | |
| "model_config": { | |
| "path": "/home/nates/musicxml/configs/model_200m_tier1.yaml", | |
| "sha256": "715bfb27addb075a7d1fb0e43fabb73886df1923d0c77a20b901d178d42fd8b5" | |
| }, | |
| "tokenizer": { | |
| "root": "/home/nates/musicxml/artifacts/tokenizer_v2_tier1", | |
| "files": [ | |
| "bpe.json", | |
| "vocab.json" | |
| ], | |
| "sha256": "e354a24a066e5060f0c21b9f516c8f49d32f190852563ffb354c450bdac40885" | |
| }, | |
| "dataset": { | |
| "root": "/home/nates/musicxml/data/shards/sft_tier1", | |
| "manifest": "/home/nates/musicxml/data/shards/sft_tier1/manifest.json", | |
| "manifest_sha256": "9de77cf72070b8ca22b75243bed59157404deb04449ff2aaffe528b9954ae8e8", | |
| "split_descriptor_sha256": "7e73c42cc3f52eef9e8f4ccaf6a81ce3a387f125ad487f6fa5c891eaa5d4b32b", | |
| "splits": { | |
| "train": { | |
| "rows": 97157, | |
| "seq_len": 2048, | |
| "shards": [ | |
| { | |
| "tokens": "train/tokens_0000.npy", | |
| "mask": "train/mask_0000.npy", | |
| "rows": 51200 | |
| }, | |
| { | |
| "tokens": "train/tokens_0001.npy", | |
| "mask": "train/mask_0001.npy", | |
| "rows": 45957 | |
| } | |
| ] | |
| }, | |
| "validation": { | |
| "rows": 1258, | |
| "seq_len": 2048, | |
| "shards": [ | |
| { | |
| "tokens": "validation/tokens_0000.npy", | |
| "mask": "validation/mask_0000.npy", | |
| "rows": 1258 | |
| } | |
| ] | |
| } | |
| } | |
| }, | |
| "environment": { | |
| "python": "3.10.12 (main, Aug 31 2026, 10:18:17) [GCC 11.4.0]", | |
| "platform": "Linux-6.8.0-1066-gcp-x86_64-with-glibc2.35", | |
| "pytorch": "2.9.1+cu129", | |
| "numpy": "2.2.6", | |
| "cuda_runtime": "12.9", | |
| "cudnn": 91002 | |
| }, | |
| "gpu": { | |
| "name": "NVIDIA RTX PRO 6000 Blackwell Server Edition", | |
| "total_memory_bytes": 101973819392, | |
| "capability": [ | |
| 12, | |
| 0 | |
| ], | |
| "driver": "580.178.04" | |
| }, | |
| "seed": 20260910 | |
| }, | |
| "params_total": 201398272, | |
| "params_trainable": 201398272, | |
| "start_step": 0, | |
| "steps_completed": 760, | |
| "elapsed_seconds_this_invocation": 2683.6809680461884, | |
| "peak_vram_gb": 45.785115648, | |
| "loss_curve": [ | |
| [ | |
| 0, | |
| 0.2846265509724617 | |
| ], | |
| [ | |
| 20, | |
| 0.2832777798175812 | |
| ], | |
| [ | |
| 40, | |
| 0.2576543968170881 | |
| ], | |
| [ | |
| 60, | |
| 0.2751853410154581 | |
| ], | |
| [ | |
| 80, | |
| 0.27040398120880127 | |
| ], | |
| [ | |
| 100, | |
| 0.23810604959726334 | |
| ], | |
| [ | |
| 120, | |
| 0.2701740376651287 | |
| ], | |
| [ | |
| 140, | |
| 0.264717735350132 | |
| ], | |
| [ | |
| 160, | |
| 0.23320921510457993 | |
| ], | |
| [ | |
| 180, | |
| 0.2678709514439106 | |
| ], | |
| [ | |
| 200, | |
| 0.24964266642928123 | |
| ], | |
| [ | |
| 220, | |
| 0.26365517638623714 | |
| ], | |
| [ | |
| 240, | |
| 0.24836344085633755 | |
| ], | |
| [ | |
| 260, | |
| 0.23373913019895554 | |
| ], | |
| [ | |
| 280, | |
| 0.2419311013072729 | |
| ], | |
| [ | |
| 300, | |
| 0.2591604311019182 | |
| ], | |
| [ | |
| 320, | |
| 0.25108602456748486 | |
| ], | |
| [ | |
| 340, | |
| 0.22293414361774921 | |
| ], | |
| [ | |
| 360, | |
| 0.23284843936562538 | |
| ], | |
| [ | |
| 380, | |
| 0.2620360627770424 | |
| ], | |
| [ | |
| 400, | |
| 0.23341848328709602 | |
| ], | |
| [ | |
| 420, | |
| 0.2336137406527996 | |
| ], | |
| [ | |
| 440, | |
| 0.22174756787717342 | |
| ], | |
| [ | |
| 460, | |
| 0.20786287263035774 | |
| ], | |
| [ | |
| 480, | |
| 0.23133441992104053 | |
| ], | |
| [ | |
| 500, | |
| 0.24537247605621815 | |
| ], | |
| [ | |
| 520, | |
| 0.22828913666307926 | |
| ], | |
| [ | |
| 540, | |
| 0.24769587628543377 | |
| ], | |
| [ | |
| 560, | |
| 0.24742664769291878 | |
| ], | |
| [ | |
| 580, | |
| 0.2207590974867344 | |
| ], | |
| [ | |
| 600, | |
| 0.2361213807016611 | |
| ], | |
| [ | |
| 620, | |
| 0.22333898954093456 | |
| ], | |
| [ | |
| 640, | |
| 0.2485098633915186 | |
| ], | |
| [ | |
| 660, | |
| 0.24824612773954868 | |
| ], | |
| [ | |
| 680, | |
| 0.24885406717658043 | |
| ], | |
| [ | |
| 700, | |
| 0.2211875542998314 | |
| ], | |
| [ | |
| 720, | |
| 0.20957332104444504 | |
| ], | |
| [ | |
| 740, | |
| 0.2173500619828701 | |
| ], | |
| [ | |
| 759, | |
| 0.23597008362412453 | |
| ] | |
| ], | |
| "val_losses": [ | |
| [ | |
| 99, | |
| 0.31183503886185754 | |
| ], | |
| [ | |
| 199, | |
| 0.31025536271044013 | |
| ], | |
| [ | |
| 299, | |
| 0.3087775707864169 | |
| ], | |
| [ | |
| 399, | |
| 0.3122648150469459 | |
| ], | |
| [ | |
| 499, | |
| 0.31319335526241254 | |
| ], | |
| [ | |
| 599, | |
| 0.31266183949211407 | |
| ], | |
| [ | |
| 699, | |
| 0.31301069269288756 | |
| ], | |
| [ | |
| 759, | |
| 0.3136875436344451 | |
| ] | |
| ], | |
| "best_val_loss": 0.3087775707864169, | |
| "best_step": 299, | |
| "micro_batch": 32, | |
| "grad_accum": 8, | |
| "seq_len": 2048, | |
| "tokens_per_step_max": 524032, | |
| "tokens_seen": 199562135, | |
| "resume_from": null, | |
| "best_checkpoint": "/home/nates/musicxml/artifacts/model_200m_sft_tier1/ckpt_best.pt", | |
| "latest_checkpoint": "/home/nates/musicxml/artifacts/model_200m_sft_tier1/ckpt_latest.pt", | |
| "precision": "torchao_fp8", | |
| "te_recipe": null, | |
| "throughput_toks_per_s": [ | |
| [ | |
| 0, | |
| 2934.2845468938017 | |
| ], | |
| [ | |
| 20, | |
| 79685.4051552035 | |
| ], | |
| [ | |
| 40, | |
| 80436.13652294016 | |
| ], | |
| [ | |
| 60, | |
| 81103.02826918027 | |
| ], | |
| [ | |
| 80, | |
| 80978.39934958126 | |
| ], | |
| [ | |
| 100, | |
| 61124.23692212198 | |
| ], | |
| [ | |
| 120, | |
| 80949.35429477402 | |
| ], | |
| [ | |
| 140, | |
| 81100.49637965474 | |
| ], | |
| [ | |
| 160, | |
| 81782.3154442603 | |
| ], | |
| [ | |
| 180, | |
| 81352.39963476159 | |
| ], | |
| [ | |
| 200, | |
| 61476.625790909646 | |
| ], | |
| [ | |
| 220, | |
| 79335.2938181554 | |
| ], | |
| [ | |
| 240, | |
| 81202.6368181386 | |
| ], | |
| [ | |
| 260, | |
| 81061.85491368544 | |
| ], | |
| [ | |
| 280, | |
| 81568.15493084869 | |
| ], | |
| [ | |
| 300, | |
| 61402.92680827696 | |
| ], | |
| [ | |
| 320, | |
| 81593.74368636096 | |
| ], | |
| [ | |
| 340, | |
| 79963.94063424174 | |
| ], | |
| [ | |
| 360, | |
| 80044.8911001716 | |
| ], | |
| [ | |
| 380, | |
| 79824.95694230743 | |
| ], | |
| [ | |
| 400, | |
| 68050.27872487987 | |
| ], | |
| [ | |
| 420, | |
| 79631.7820692823 | |
| ], | |
| [ | |
| 440, | |
| 80400.248119905 | |
| ], | |
| [ | |
| 460, | |
| 79856.98116091211 | |
| ], | |
| [ | |
| 480, | |
| 81333.78704798948 | |
| ], | |
| [ | |
| 500, | |
| 68311.21101458241 | |
| ], | |
| [ | |
| 520, | |
| 79491.89788235497 | |
| ], | |
| [ | |
| 540, | |
| 80735.1329593898 | |
| ], | |
| [ | |
| 560, | |
| 83276.14449517097 | |
| ], | |
| [ | |
| 580, | |
| 81067.41114257302 | |
| ], | |
| [ | |
| 600, | |
| 69974.22693335732 | |
| ], | |
| [ | |
| 620, | |
| 82367.9521560698 | |
| ], | |
| [ | |
| 640, | |
| 80877.23511417964 | |
| ], | |
| [ | |
| 660, | |
| 81774.84449776652 | |
| ], | |
| [ | |
| 680, | |
| 80270.33452565636 | |
| ], | |
| [ | |
| 700, | |
| 69265.66820467077 | |
| ], | |
| [ | |
| 720, | |
| 78387.42052580223 | |
| ], | |
| [ | |
| 740, | |
| 80889.83415720014 | |
| ], | |
| [ | |
| 759, | |
| 81170.87283314017 | |
| ] | |
| ], | |
| "ppl_curve": [ | |
| [ | |
| 99, | |
| 1.3659293491834388 | |
| ], | |
| [ | |
| 199, | |
| 1.3637733265231424 | |
| ], | |
| [ | |
| 299, | |
| 1.3617594417328527 | |
| ], | |
| [ | |
| 399, | |
| 1.36651651925505 | |
| ], | |
| [ | |
| 499, | |
| 1.3677859740769955 | |
| ], | |
| [ | |
| 599, | |
| 1.367059167433088 | |
| ], | |
| [ | |
| 699, | |
| 1.3675361535935309 | |
| ], | |
| [ | |
| 759, | |
| 1.368462085050144 | |
| ] | |
| ] | |
| } |