tsuzumi-7b

Version: 2

NTT Data•Last updated May 2026

Models from Microsoft, Partners, and Community

Models from Microsoft, Partners, and Community models are a select portfolio of curated models both general-purpose and niche models across diverse scenarios by developed by Microsoft teams, partners, and community contributors

Managed by Microsoft: Purchase and manage models directly through Azure with a single license, world class support and enterprise grade Azure infrastructure
Validated by providers: Each model is validated and maintained by its respective provider, with Azure offering integration and deployment guidance.
Innovation and agility: Combines Microsoft research models with rapid, community-driven advancements.
Seamless Azure integration: Standard Microsoft Foundry experience, with support managed by the model provider.
Flexible deployment: Deployable as Managed Compute or Serverless API, based on provider preference.

Learn more about models from Microsoft, Partners, and Community

Key capabilities

About this model

tsuzumi is a lightweight large language model developed by NTT. tsuzumi is designed to handle both Japanese and English with high efficiency. tsuzumi was trained to follow instructions in both Japanese and English.

Key model capabilities

Instruction following in both Japanese and English
Text generation
Auto-regressive language modeling with transformer architecture
Supervised fine-tuning (SFT) for response learning

Model	Size	Japanese MT-bench; turn1? Japanese / English
Model	Size	writing	stem	humanities	roleplay	extraction	coding	math	reasoning
tsuzumi-7B	7B	8.6 / 8.2	7.6 / 7.1	8.45 / 8.2	6.3 / 6.05	5.6 / 2.9	2.3 / 2.2	1.1 / 1.1	2.1 / 4.3

https://github.com/Stability-AI/FastChat/tree/jp-stable/fastchat/llm_judge ?Evaluated in 1 turn only

Use cases

See Responsible AI for additional considerations for responsible use.

Key use cases

The provider has not supplied this information.

Out of scope use cases

The provider has not supplied this information.

Pricing

Pricing is based on a number of factors, including deployment type and tokens used. See pricing details here.

Technical specs

tsuzumi is an auto-regressive language optimized transformer. The tuned versions use supervised fine-tuning (SFT).

Name	Training Data	Params	Content Length	GQA	Tokens
tsuzumi-7b	A mix of publicly available online and private data	7B	8k	?	1.4T

Training cut-off date

tsuzumi was trained until 2024/08; Knowledge cutoff is 2024/05. The pretraining data has a cutoff of May 2024.

Training time

tsuzumi was trained until August 2024.

Input formats

Models input text only.

Output formats

Models generate text only.

Supported languages

tsuzumi is designed to handle both Japanese and English.

Sample JSON response

{
  "output": "Linux?????????????????????????????????????????????????????????????????????\n\n1. ???????: Linux??????????????????????????????????????????????????????????????????? ?????????????????????????????OS???????????????\n\n2. ??????: Linux??????????????????????? ?????????????????????????????????????????????????????????????????????\n\n3. ???: Linux?? ?????????????????????????????????????????????????????????????????????????????PC??? ????????????????????????????·????????????\n\n4. ??????: Linux OS???????????????????????????????????????????????????????????????????????????????????????????"
}

Model architecture

tsuzumi is an auto-regressive language optimized transformer. The tuned versions use supervised fine-tuning (SFT).

Long context

The provider has not supplied this information.

Optimizing model performance

Basic parameters:

name	Defaults to	Explanation
temperature	0.15	Controls randomness in the model. Lower values will make the model more deterministic and higher values will make the model more random.
max_tokens	4096	The maximum number of tokens to generate.
top_p	1.0	The cumulative probability of parameter highest probability vocabulary tokens to keep for nucleus sampling.
frequency_penalty	0.0	Positive values penalize new tokens based on their existing frequency in the text so far, decreasing the model's likelihood to repeat the same line verbatim.
logit_bias	null	Modify the likelihood of specified tokens appearing in the completion.
logprobs	false	Whether to return log probabilities of the output tokens or not.
top_logprobs	0	An integer between 0 and 20 specifying the number of most likely tokens to return at each token position, each with an associated log probability.
n	1	The number of generated response variations.
presence_penalty	0.0	Positive values penalize new tokens based on whether they appear in the text so far, increasing the model's likelihood to talk about new topics.
stop	null	Specification for stop words.
stream	false	Whether response is returned in partial message deltas.

Advanced Parameters – To specify, set "extra-params: allow" in your HTTP request header:

name	Defaults to	explanation
min_tokens	0	The minimum number of tokens to generate.
top_k	-1 (no filter)	The number of highest probability vocabulary tokens to keep for top-k-filtering.
repetition_penalty	1.0	The weight of penalty for repeated phrases. Higher values will suppress repeating similar phrases.
length_penalty	1.0	Penalizes sequences based on their length.

Additional assets

The provider has not supplied this information.

Training disclosure

Training, testing and validation

A mix of publicly available online and private data was used for training. The pretraining data has a cutoff of May 2024.

Distribution

Distribution channels

Available Regions: US EAST 2

More information

Download (?????????) - Terms of Use Link to Download page (?tsuzumi on Azure MaaS????????) - User Guide

Responsible AI considerations

Safety techniques

The provider has not supplied this information.

Safety evaluations

The provider has not supplied this information.

Known limitations

The provider has not supplied this information.

Acceptable use

Acceptable use policy

The provider has not supplied this information.

Quality and performance evaluations

Source: NTT Data In this section, we report the results for the tsuzumi models on Japanese standard benchmarks.

Model	Size	Japanese MT-bench; turn1※　日本語 / 英語
Model	Size	writing	stem	humanities	roleplay	extraction	coding	math	reasoning
tsuzumi-7B	7B	8.6 / 8.2	7.6 / 7.1	8.45 / 8.2	6.3 / 6.05	5.6 / 2.9	2.3 / 2.2	1.1 / 1.1	2.1 / 4.3

https://github.com/Stability-AI/FastChat/tree/jp-stable/fastchat/llm_judge 　※1ターンでの評価を採用

Benchmarking methodology

Source: NTT Data The provider has not supplied this information.

Public data summary

Source: NTT Data The provider has not supplied this information.

Model Specifications

Last UpdatedMay 2026

Input TypeText

Output TypeText

ProviderNTT Data

Languages1 Language

Quick Start