Japanese-focused chat use
Hub data records conversational text-generation with a Japanese tag, and the model card describes a Japanese-focused small chat model for embedded environments.
Open Source Model Profile · 2121-8
TinySlime-1.1B-Chat-v1.0
TinySlime-1.1B-Chat-v1.0 is a 1.1B-parameter Japanese-focused Llama chat model from 2121-8. The model card documents Axolotl tuning and synthetic Mixtral-derived data.
- Publisher
- 2121-8
- Task
- text-generation
- Model type
- llama
- License
- apache-2.0
- Library
- transformers
- Publication status
- Accepted · not indexed
Model overview
TinySlime-1.1B-Chat-v1.0 is published by 2121-8 as a Llama text-generation chat model. The captured configuration identifies LlamaForCausalLM with model type llama, and Safetensors metadata reports 1,100,048,384 parameters. According to the model card, it is a Japanese-focused small model tuned for embedded use.
Recorded capabilities
1.1B Llama record with Axolotl
Captured configuration records LlamaForCausalLM with about 1.1B parameters, and the model card states full tuning was done with Axolotl.
Publisher-reported Japanese benchmarks
According to the model card, the JP Harness table reports TinySlime-Chat average 53.50 across JCommonsenseQA, JNLI, MARC-ja, JSQuAD, and other Japanese tasks.
Documented Transformers loading
According to the model card, loading uses AutoModelForCausalLM and AutoTokenizer with a chat template and generated-text decoding example.
Use cases in the source record
- Japanese conversational text generation on small or embedded-oriented setups suggested by the card's smartphone and Jetson note.
- Japanese-task comparison that treats the card's JP Harness table as publisher-reported rather than verified.
Limitations and unknowns
- No context-window value was extracted from this record.
- Reported Japanese benchmark figures are publisher claims from the model card and were not independently verified.
- Base-model lineage wording in the captured card text is inconsistent, so the exact parent checkpoint remains unclear.
- Provider state is historical snapshot data and should be refreshed before being presented as current.
Source and provenance
Source: 2121-8/TinySlime-1.1B-Chat-v1.0
Captured: Unknown. Processed: 2026-09-07T19:34:28.088449+00:00.
TinySlime-1.1B-Chat-v1.0 TinySlime は日本語に特化した小規模言語モデルです。 モデルの一覧 TinySlime-1.1B-v1.0 TinySlime-1.1B-Chat-v1.0 このモデルのフルチューニングは、 Axolotl を使用して行われました。 モデル概要 TinySlime-1.1B-Chat-v1.0 は、TinySlime-1.1B-Chat-v1.0 をベースに、Mixtral-8x7B-Instruct-v0.1 が作成した合成データを使用して微調整されたチャットモデルです。 このモデルは、スマートフォンや NVIDIA Jetson などの組み込み環境で動かすことを想定して作成されました。 ベンチマーク(JP Language Model Evaluation Harness) model name JCommonsenseQA (3-shot) JNLI (3-shot) MARC-ja (0-shot) JSQuAD (2-shot) jaqket-v2 (1-shot) xlsum (1-shot) xwinograd (0-shot) mgsm (5-shot) AVERAGE lkarasu-1.1B 24.84 34.78 50.48 51.33 18.95 9.21 62.15 2.80 31.82 TinySlime-Chat 66.49 64.36 92.68 84.60 42.14…
F001F002F003F004F005F006F007F009F010F011F012F013F014