Skip to page content
SAVRN
Search Plan a project

Open-weight model

arka-v5-8k-mixed-sft-54m-v1

by AKRP Abhayn01/arka-v5-8k-mixed-sft-54m-v1

arka-v5-8k-mixed-sft-54m-v1 is an open-weight model from AKRP. Its published files total 5.2 GB.

Parameters—
Context—
Weights5.2 GB
License—
AccessOpen weights
Monthly Downloads—

Model Card

The publisher has not written a card for this model.

Identity and Version

Repository
Abhayn01/arka-v5-8k-mixed-sft-54m-v1
Publisher
AKRP
Task
Not stated by the source
Modality
Other
Library
Not stated by the source
Parameters
Not stated by the source
Languages
Not stated by the source
Revision
b801d9ebb1d654046df8ce08b17848dfa95c3183
First published
2026-09-27
Last updated
2026-09-28

Files and Weights

78 files, 5.2 GB in total. The weights are 10 files totalling 5.2 GB in pt, safetensors.

Weights10 files · 5.2 GB
Configuration20 files · 17.2 KB
Tokenizer8 files · 14.2 MB
Other39 files · 198.1 KB
Repository1 file · 1.5 KB
Every file
FileTypeSizeSHA-256
assistant_alignment_final_packed_v2/model.safetensorsWeights519.8 MB 451d300116a7
assistant_alignment_final_packed_v2/optimizer.ptWeights1.0 GB 240d79a7906b
assistant_alignment_final_packed_v2/scaler.ptWeights1.4 KB aed8c88f0276
latest/model.safetensorsWeights519.8 MB e55df00b714f
latest/optimizer.ptWeights1.0 GB 400492a6c13e
latest/scaler.ptWeights1.4 KB 7ca5e0e1873a
mixed_sft_final/model.safetensorsWeights519.8 MB 5a93ea277ba4
qa_instruction_rag_10m_final/model.safetensorsWeights519.8 MB e55df00b714f
qa_instruction_rag_10m_final/optimizer.ptWeights1.0 GB 400492a6c13e
qa_instruction_rag_10m_final/scaler.ptWeights1.4 KB 7ca5e0e1873a
assistant_alignment_final_packed_v2/alignment_dataset_manifest.jsonConfiguration772 B —
assistant_alignment_final_packed_v2/config.jsonConfiguration1.4 KB —
assistant_alignment_final_packed_v2/generation_config.jsonConfiguration118 B —
assistant_alignment_final_packed_v2/trainer_state.jsonConfiguration698 B —
dataset_manifest.jsonConfiguration1.5 KB —
latest/alignment_dataset_manifest.jsonConfiguration772 B —
latest/config.jsonConfiguration1.4 KB —
latest/dataset_manifest.jsonConfiguration1.3 KB —
latest/dataset_revisions.jsonConfiguration275 B —
latest/generation_config.jsonConfiguration118 B —
latest/trainer_state.jsonConfiguration938 B —
mixed_sft_final/config.jsonConfiguration1.4 KB —
mixed_sft_final/dataset_manifest.jsonConfiguration1.5 KB —
mixed_sft_final/generation_config.jsonConfiguration118 B —
mixed_sft_final/trainer_state.jsonConfiguration817 B —
qa_instruction_rag_10m_final/config.jsonConfiguration1.4 KB —
qa_instruction_rag_10m_final/dataset_manifest.jsonConfiguration1.3 KB —
qa_instruction_rag_10m_final/dataset_revisions.jsonConfiguration275 B —
qa_instruction_rag_10m_final/generation_config.jsonConfiguration118 B —
qa_instruction_rag_10m_final/trainer_state.jsonConfiguration938 B —
alignment_reports/response_step_00000.txtOther6.5 KB —
alignment_reports/response_step_00050.txtOther5.2 KB —
alignment_reports_packed_v2/response_step_0000.txtOther6.3 KB —
alignment_reports_packed_v2/response_step_0020.txtOther4.1 KB —
alignment_reports_packed_v2/response_step_0040.txtOther3.9 KB —
alignment_reports_packed_v2/response_step_0060.txtOther3.9 KB —
alignment_reports_packed_v2/response_step_0080.txtOther3.8 KB —
alignment_reports_packed_v2/response_step_0100.txtOther3.8 KB —
alignment_reports_packed_v2/response_step_0120.txtOther3.8 KB —
alignment_reports_packed_v2/response_step_0140.txtOther3.8 KB —
alignment_reports_packed_v2/response_step_0160.txtOther3.8 KB —
alignment_reports_packed_v2/response_step_0180.txtOther3.8 KB —
alignment_reports_packed_v2/response_step_0187.txtOther3.8 KB —
assistant_alignment_final_packed_v2/chat_template.jinjaOther343 B —
latest/chat_template.jinjaOther343 B —
mixed_sft_final/chat_template.jinjaOther343 B —
qa_instruction_rag_10m_final/chat_template.jinjaOther343 B —
response_tests/step_000000.txtOther7.2 KB —
response_tests/step_000100.txtOther5.6 KB —
response_tests/step_000200.txtOther6.3 KB —
response_tests/step_000300.txtOther6.6 KB —
response_tests/step_000400.txtOther6.6 KB —
response_tests/step_000500.txtOther6.9 KB —
response_tests/step_000600.txtOther6.0 KB —
response_tests/step_000700.txtOther6.4 KB —
response_tests/step_000800.txtOther6.3 KB —
response_tests/step_000900.txtOther6.6 KB —
response_tests/step_001000.txtOther6.3 KB —
response_tests/step_001100.txtOther6.7 KB —
response_tests/step_001200.txtOther5.8 KB —
response_tests/step_001300.txtOther6.8 KB —
response_tests/step_001400.txtOther6.2 KB —
response_tests/step_001500.txtOther6.1 KB —
response_tests/step_001600.txtOther5.8 KB —
response_tests/step_001700.txtOther7.2 KB —
response_tests/step_001800.txtOther6.0 KB —
response_tests/step_001900.txtOther6.6 KB —
response_tests/step_002000.txtOther6.1 KB —
response_tests/step_002038.txtOther6.1 KB —
.gitattributesRepository1.5 KB —
assistant_alignment_final_packed_v2/tokenizer.jsonTokenizer3.6 MB —
assistant_alignment_final_packed_v2/tokenizer_config.jsonTokenizer433 B —
latest/tokenizer.jsonTokenizer3.6 MB —
latest/tokenizer_config.jsonTokenizer433 B —
mixed_sft_final/tokenizer.jsonTokenizer3.6 MB —
mixed_sft_final/tokenizer_config.jsonTokenizer433 B —
qa_instruction_rag_10m_final/tokenizer.jsonTokenizer3.6 MB —
qa_instruction_rag_10m_final/tokenizer_config.jsonTokenizer433 B —

License and Download

License
Not stated by the source
Access
Open weights, no gate
Download size
5.2 GB
Download from AKRP

Released by AKRP through its official repository on Hugging Face.

Memory Requirements

PrecisionWeights in memory
As published5.2 GB

Weights only, from the published parameter count; the key-value cache and runtime add to this.