MP_political_MIL_classifier is an open-weight model from Oriane Peter. It has 434M parameters and a 512-token context. At 16-bit it needs about 1 GB of GPU memory, which fits on 1x MI300X from $1.85 an hour, at the lowest prices in the SAVRN Index. It draws 819 downloads a month.
This model is a fine-tuned version of on an unknown dataset. The following hyperparameters were used during training: - learningrate: 1e-05 - trainbatchsize: 2 - evalbatchsize: 4 - gradientaccumulationsteps: 4 - totaltrainbatchsize: 8 - lrschedulertype…
Runs On
What it takes to serve MP_political_MIL_classifier (434M parameters): the memory its weights need at each precision, and the cheapest way to rent enough data-center GPUs to hold them.
| Precision | Weights | Memory needed | Cheapest setup | Per hour | Also fits |
|---|---|---|---|---|---|
| 16-bit | 0.9 GB | 1.0 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 8-bit | 0.4 GB | 0.5 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 4-bit | 0.2 GB | 0.3 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
Memory is the weights at that precision plus 20% for the runtime and a short context; a long context needs more. Prices are the lowest on-demand hourly rates in the SAVRN Index, read Sep 18, 2026.
MP_political_MIL_classifier on every accelerator the SAVRN Index prices, at every precision
Model Card
This model is a fine-tuned version of on an unknown dataset. The following hyperparameters were used during training: - learningrate: 1e-05 - trainbatchsize: 2 - evalbatchsize: 4 - gradientaccumulationsteps: 4 - totaltrainbatchsize: 8 - lrschedulertype: cosine - lrschedulerwarmupsteps: 0.1 - numepochs: 50 - Transformers 5.12.1 - Pytorch 2.5.1+cu121 - Datasets 5.0.1 - Tokenizers 0.22.2
Excerpt from the card by Oriane Peter.
Configuration
- Architecture
- DebertaLWAN
- Context length (tokens)
- 512
- Model type
- deberta_LWAN
Identity and Version
- Repository
- orpe42/MP_political_MIL_classifier
- Publisher
- Oriane Peter
- Task
- Not stated by the source
- Modality
- Other
- Library
- transformers
- Parameters
- 434M parameters
- Languages
- Not stated by the source
- Revision
- 8810fbe4bcfe3482208207758affbca248516ea3
- First published
- 2026-08-10
- Last updated
- 2026-09-18
Files and Weights
79 files, 3.5 GB in total. The weights are 4 files totalling 3.5 GB in bin, safetensors.
Every file
| File | Type | Size | SHA-256 |
|---|---|---|---|
| best_model/model.safetensors | Weights | 1.7 GB | dec3f0a3606c |
| best_model/training_args.bin | Weights | 4.9 KB | bdffe459167b |
| model.safetensors | Weights | 1.7 GB | ff57fbf8ce4c |
| training_args.bin | Weights | 4.9 KB | 59ba34aa2116 |
| best_model/config.json | Configuration | 4.5 KB | — |
| config.json | Configuration | 4.5 KB | — |
| README.md | Documentation | 1.2 KB | — |
| evaluation_results.txt | Other | 15.6 KB | — |
| runs/Aug10_18-18-27_erc-hpc-comp240/events.out.tfevents.1786382307.erc-hpc-comp240.3508104.0 | Other | 11.5 KB | f5b8b9b5b549 |
| runs/Aug10_19-00-57_erc-hpc-comp032/events.out.tfevents.1786384857.erc-hpc-comp032.145948.0 | Other | 525.7 KB | 0c1793783bdb |
| runs/Aug11_08-46-10_erc-hpc-comp033/events.out.tfevents.1786434370.erc-hpc-comp033.3501889.0 | Other | 8.9 KB | 22dd795f13bb |
| runs/Aug11_09-18-13_erc-hpc-comp248/events.out.tfevents.1786436293.erc-hpc-comp248.2607868.0 | Other | 9.9 KB | 8ff713151f46 |
| runs/Aug11_10-02-01_erc-hpc-comp040/events.out.tfevents.1786438921.erc-hpc-comp040.1724311.0 | Other | 9.3 KB | 38f774b0e0f3 |
| runs/Aug11_10-10-47_erc-hpc-comp036/events.out.tfevents.1786439447.erc-hpc-comp036.489496.0 | Other | 9.9 KB | 5db00be22ef3 |
| runs/Aug11_10-42-03_erc-hpc-comp040/events.out.tfevents.1786441323.erc-hpc-comp040.1726396.0 | Other | 8.9 KB | 18b35ffb0632 |
| runs/Aug11_10-51-01_erc-hpc-comp040/events.out.tfevents.1786441861.erc-hpc-comp040.1727375.0 | Other | 8.9 KB | a06dcd644454 |
| runs/Aug11_11-05-13_erc-hpc-comp040/events.out.tfevents.1786442713.erc-hpc-comp040.1727846.0 | Other | 8.9 KB | 2bc5ddaf5d7f |
| runs/Aug11_11-10-01_erc-hpc-comp040/events.out.tfevents.1786443001.erc-hpc-comp040.1728979.0 | Other | 8.9 KB | b7f5c5d1d3a0 |
| runs/Aug11_11-14-13_erc-hpc-comp040/events.out.tfevents.1786443253.erc-hpc-comp040.1729192.0 | Other | 8.9 KB | cacf8567e4ea |
| runs/Aug11_11-16-19_erc-hpc-comp040/events.out.tfevents.1786443379.erc-hpc-comp040.1729404.0 | Other | 8.9 KB | a3b2f412d712 |
| runs/Aug11_11-18-28_erc-hpc-comp040/events.out.tfevents.1786443508.erc-hpc-comp040.1729607.0 | Other | 8.9 KB | 833901e3cf9f |
| runs/Aug11_11-35-17_erc-hpc-comp040/events.out.tfevents.1786444517.erc-hpc-comp040.1731370.0 | Other | 8.9 KB | d0d99c6c6e2b |
| runs/Aug11_11-39-39_erc-hpc-comp040/events.out.tfevents.1786444779.erc-hpc-comp040.1731577.0 | Other | 77.1 KB | 45b71db8c1d0 |
| runs/Aug11_11-39-39_erc-hpc-comp040/events.out.tfevents.1786466892.erc-hpc-comp040.1731577.1 | Other | 1.8 KB | 1c1780a8271c |
| runs/Aug11_11-51-57_erc-hpc-vm065/events.out.tfevents.1786445517.erc-hpc-vm065.63422.0 | Other | 11.6 KB | 07e4160d4e34 |
| runs/Aug11_12-37-33_erc-hpc-comp248/events.out.tfevents.1786448253.erc-hpc-comp248.2615863.0 | Other | 77.1 KB | ec1eac3d047a |
| runs/Aug11_12-37-33_erc-hpc-comp248/events.out.tfevents.1786463313.erc-hpc-comp248.2615863.1 | Other | 1.8 KB | d6045ab2ee70 |
| runs/Aug12_14-01-53_erc-hpc-comp053/events.out.tfevents.1786539713.erc-hpc-comp053.456620.0 | Other | 39.0 KB | 15835c5a1407 |
| runs/Aug12_23-28-27_erc-hpc-comp038/events.out.tfevents.1786573707.erc-hpc-comp038.1039659.0 | Other | 121.6 KB | 6d3abead6bf4 |
| runs/Aug12_23-28-27_erc-hpc-comp038/events.out.tfevents.1786600750.erc-hpc-comp038.1039659.1 | Other | 1.8 KB | ba65f56f18f8 |
| runs/Aug18_10-17-36_erc-hpc-comp222/events.out.tfevents.1787044656.erc-hpc-comp222.1438054.0 | Other | 3.6 KB | 6245d956640c |
| runs/Aug18_10-23-32_erc-hpc-comp222/events.out.tfevents.1787045012.erc-hpc-comp222.1438115.0 | Other | 3.6 KB | 5c958f5221d4 |
| runs/Aug18_10-26-19_erc-hpc-comp222/events.out.tfevents.1787045179.erc-hpc-comp222.1438179.0 | Other | 3.6 KB | 95f242e7b56b |
| runs/Aug18_10-31-46_erc-hpc-comp222/events.out.tfevents.1787045506.erc-hpc-comp222.1438328.0 | Other | 3.6 KB | d69eb23eace2 |
| runs/Aug18_10-36-42_erc-hpc-comp222/events.out.tfevents.1787045802.erc-hpc-comp222.1438363.0 | Other | 3.6 KB | 17f891c541c6 |
| runs/Aug18_10-38-09_erc-hpc-comp222/events.out.tfevents.1787045889.erc-hpc-comp222.1438385.0 | Other | 3.6 KB | 171450b73ea8 |
| runs/Aug18_10-38-46_erc-hpc-comp222/events.out.tfevents.1787045926.erc-hpc-comp222.1438422.0 | Other | 8.1 KB | 374ca24719a1 |
| runs/Aug18_10-43-05_erc-hpc-comp222/events.out.tfevents.1787046185.erc-hpc-comp222.1438480.0 | Other | 16.3 KB | 49db9b854bae |
| runs/Aug18_11-45-34_erc-hpc-comp222/events.out.tfevents.1787049934.erc-hpc-comp222.1440173.0 | Other | 8.1 KB | 627fc58d6ef1 |
| runs/Aug18_11-47-24_erc-hpc-comp222/events.out.tfevents.1787050044.erc-hpc-comp222.1440220.0 | Other | 8.1 KB | 4cba0a4d35d1 |
| runs/Aug18_11-58-44_erc-hpc-comp222/events.out.tfevents.1787050724.erc-hpc-comp222.1440268.0 | Other | 9.8 KB | 2b9c11cab19d |
| runs/Aug18_12-19-01_erc-hpc-comp222/events.out.tfevents.1787051941.erc-hpc-comp222.1441082.0 | Other | 15.2 KB | e6bfcf7a58d4 |
| runs/Aug18_15-19-18_erc-hpc-vm046/events.out.tfevents.1787062758.erc-hpc-vm046.452799.0 | Other | 159.1 KB | 2354f0ea6b8d |
| runs/Aug18_15-19-26_erc-hpc-comp248/events.out.tfevents.1787062766.erc-hpc-comp248.1575216.0 | Other | 60.1 KB | 5554b5274a16 |
| runs/Aug19_09-50-02_erc-hpc-vm042/events.out.tfevents.1787129402.erc-hpc-vm042.122275.0 | Other | 13.6 KB | 65ff0e6593bb |
| runs/Aug19_10-22-07_erc-hpc-comp127/events.out.tfevents.1787131327.erc-hpc-comp127.3204075.0 | Other | 63.0 KB | 1bd80ca5186c |
| runs/Aug19_10-32-16_erc-hpc-comp248/events.out.tfevents.1787131936.erc-hpc-comp248.1886642.0 | Other | 27.3 KB | dd2a15b458fc |
| runs/Aug19_10-34-31_erc-hpc-comp248/events.out.tfevents.1787132071.erc-hpc-comp248.1887268.0 | Other | 60.2 KB | 900f23f2a37b |
| runs/Aug19_12-03-39_erc-hpc-vm047/events.out.tfevents.1787137419.erc-hpc-vm047.2896772.0 | Other | 60.2 KB | 015222f21120 |
| runs/Aug19_12-22-30_erc-hpc-vm046/events.out.tfevents.1787138550.erc-hpc-vm046.476251.0 | Other | 60.2 KB | b8ae0301a0f0 |
| runs/Aug19_17-26-25_erc-hpc-comp039/events.out.tfevents.1787156785.erc-hpc-comp039.134633.0 | Other | 8.1 KB | 9d46cd70ab2c |
| runs/Aug19_17-27-30_erc-hpc-comp039/events.out.tfevents.1787156850.erc-hpc-comp039.134873.0 | Other | 8.4 KB | 488cc1712933 |
| runs/Aug19_17-36-39_erc-hpc-comp039/events.out.tfevents.1787157399.erc-hpc-comp039.135336.0 | Other | 109.1 KB | 383167ca1c5e |
| runs/Aug19_17-36-39_erc-hpc-comp039/events.out.tfevents.1787189501.erc-hpc-comp039.135336.1 | Other | 1.8 KB | 924aca80cc72 |
| runs/Aug19_18-53-53_erc-hpc-comp222/events.out.tfevents.1787162033.erc-hpc-comp222.1480327.0 | Other | 21.8 KB | 53dd89331aa0 |
| runs/Aug19_21-06-13_erc-hpc-comp036/events.out.tfevents.1787169973.erc-hpc-comp036.1595614.0 | Other | 137.6 KB | a4e8de0a0625 |
| runs/Aug19_22-10-22_erc-hpc-comp039/events.out.tfevents.1787173822.erc-hpc-comp039.142671.0 | Other | 123.7 KB | fb4147d8db05 |
| runs/Aug20_08-38-28_erc-hpc-comp039/events.out.tfevents.1787211508.erc-hpc-comp039.159253.0 | Other | 176.4 KB | 4b95119ffe73 |
| runs/Aug20_08-38-28_erc-hpc-comp039/events.out.tfevents.1787264782.erc-hpc-comp039.159253.1 | Other | 1.8 KB | c04ffec2b606 |
| runs/Aug20_08-46-05_erc-hpc-comp039/events.out.tfevents.1787211965.erc-hpc-comp039.159493.0 | Other | 279.7 KB | 59b1d5e36592 |
| runs/Aug20_10-57-48_erc-hpc-comp222/events.out.tfevents.1787219868.erc-hpc-comp222.1496469.0 | Other | 8.1 KB | 9f477ed128c0 |
| runs/Aug20_10-59-07_erc-hpc-comp222/events.out.tfevents.1787219947.erc-hpc-comp222.1496504.0 | Other | 13.6 KB | 9e32ec6e2849 |
| runs/Aug20_18-54-05_erc-hpc-comp038/events.out.tfevents.1787248445.erc-hpc-comp038.48962.0 | Other | 82.1 KB | 9a78cf059a4b |
| runs/Aug21_15-14-02_erc-hpc-comp222/events.out.tfevents.1787321642.erc-hpc-comp222.14118.0 | Other | 30.0 KB | eb24f0c2033c |
| runs/Aug21_17-47-23_erc-hpc-comp039/events.out.tfevents.1787330843.erc-hpc-comp039.216194.0 | Other | 84.9 KB | 107cb7e7805f |
| runs/Aug24_16-02-01_erc-hpc-vm045/events.out.tfevents.1787583721.erc-hpc-vm045.1046608.0 | Other | 112.7 KB | c96487fcb381 |
| runs/Aug24_17-28-27_erc-hpc-comp039/events.out.tfevents.1787588907.erc-hpc-comp039.1450230.0 | Other | 143.2 KB | 76ec4f926c09 |
| runs/Aug25_09-01-26_erc-hpc-comp248/events.out.tfevents.1787644886.erc-hpc-comp248.2198170.0 | Other | 162.7 KB | 60fa41392529 |
| runs/Sep15_22-59-35_erc-hpc-comp036/events.out.tfevents.1789509575.erc-hpc-comp036.4146001.0 | Other | 109.8 KB | 75d65fbb83fe |
| runs/Sep17_09-41-28_erc-hpc-comp036/events.out.tfevents.1789634488.erc-hpc-comp036.849349.0 | Other | 10.0 KB | 8ee95e763639 |
| runs/Sep17_11-22-40_erc-hpc-vm041/events.out.tfevents.1789640560.erc-hpc-vm041.776138.0 | Other | 204.5 KB | 62437e68f4ef |
| runs/Sep17_17-28-10_erc-hpc-comp032/events.out.tfevents.1789662490.erc-hpc-comp032.1422959.0 | Other | 76.7 KB | 9933c73eae47 |
| runs/Sep18_09-37-25_erc-hpc-comp248/events.out.tfevents.1789720645.erc-hpc-comp248.2194917.0 | Other | 32.8 KB | e0422b5c5de4 |
| .gitattributes | Repository | 1.5 KB | — |
| .npy | Repository | 576 B | a82d0e78532e |
| best_model/tokenizer.json | Tokenizer | 8.3 MB | — |
| best_model/tokenizer_config.json | Tokenizer | 538 B | — |
| tokenizer.json | Tokenizer | 8.3 MB | — |
| tokenizer_config.json | Tokenizer | 538 B | — |
License and Download
- License
- Not stated by the source
- Access
- Open weights, no gate
- Download size
- 3.5 GB
Released by Oriane Peter through its official repository on Hugging Face.
Memory Requirements
| Precision | Weights in memory |
|---|---|
| As published | 3.5 GB |
| 16-bit | 0.9 GB |
| 8-bit | 0.4 GB |
| 4-bit | 0.2 GB |
Weights only, from the published parameter count; the key-value cache and runtime add to this.
Questions About MP_political_MIL_classifier
How much GPU memory does MP_political_MIL_classifier need?
About 1 GB at 16-bit and 0.3 GB at 4-bit: the weights (434M parameters) plus a working margin. A long context needs more.
What is the cheapest GPU to run MP_political_MIL_classifier on?
At 16-bit, 1x MI300X from $1.85 an hour; at 4-bit, 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists.
What is MP_political_MIL_classifier's context length?
512 tokens, from the maximum position embeddings in its published configuration.