File size: 4,959 Bytes

---
language:
- en
library_name: transformers
pipeline_tag: any-to-any
tags:
- tachibana
- valiant
- valiant-labs
- meta
- facebook
- muse-glimmer
- muse
- glimmer
- muse-glimmer-30b
- 30b
- reasoning
- code
- code-instruct
- python
- typescript
- javascript
- java
- c++
- c
- c#
- rust
- go
- haskell
- shell
- bash
- azure
- aws
- gcp
- cloud
- scripting
- powershell
- problem-solving
- architect
- engineer
- developer
- creative
- analytical
- expert
- rationality
- conversational
- chat
- instruct
base_model: meta-models/Muse-Glimmer-30B
datasets:
- sequelbox/Tachibana4-DeepSeek-V4-Pro
license: apache-2.0
---


**[Support our open-source dataset and model releases!](https://huggingface.co/spaces/sequelbox/SupportOpenSource)**


Tachibana-Agent: [gemma-4-12B](https://huggingface.co/sequelbox/gemma-4-12B-it-Tachibana-Agent), [Qwen3.6-27B](https://huggingface.co/sequelbox/Qwen3.6-27B-Tachibana-Agent), [Muse-Glimmer-30B](https://huggingface.co/sequelbox/Muse-Glimmer-30B-Tachibana-Agent)


Tachibana-Agent is a Muse Glimmer agentic coding finetune, trained on the [Tachibana 4](https://huggingface.co/datasets/sequelbox/Tachibana4-DeepSeek-V4-Pro) dataset.
- Questions prioritize real-world, challenging agentic coding tasks across a variety of programming languages and topics. Synthetic prompts utilize a variety of personas, experience levels, and styles of communication to maximize real-world flexibility and usability.
- Areas of focus include back-end and front-end development, systems programming, distributed systems, performance optimization, data structures, databases and data engineering, game and mobile development, security engineering, compiler design, custom tooling, task automation, practical bugfixes, and more!
- A wide variety of emphasized languages improves development capability: Python, C, C++, C#, Go, TypeScript, Java, JavaScript, Rust, Haskell, SQL, Shell, R, Ruby, assembly code, and more!


## Prompting Guide
Tachibana-Agent uses the [Muse Glimmer](https://huggingface.co/meta-models/Muse-Glimmer-30B) prompt format and the following recommended general structure:

1) Start the prompt with your primary query
2) Include reference information after the primary query, using subheaders; documentation should follow "Documentation:\n\n", a stack trace following "Stack Trace:\n\n", etc for logs, schemas, specs, etc.
3) Attached files for the agent go at the end, with each file surrounded by file tags: <file path="myStuff/myRepo/myFirstFile.scala" language=Scala"> and \</file>

Adherence to the specific format above is not required, but reflects the structure of the training data.

Example inference script to get started:

```python
from transformers import AutoProcessor, AutoModelForMultimodalLM

MODEL_ID = "sequelbox/Muse-Glimmer-30B-Tachibana-Agent"

# Load model
processor = AutoProcessor.from_pretrained(MODEL_ID)
model = AutoModelForMultimodalLM.from_pretrained(
    MODEL_ID,
    dtype="auto",
    device_map="auto"
)

# Prompt
prompt = "Implement CQRS for network appliance config management.\n\nRequirements:\n- Write side: 200 commands/sec, 4 command handlers, SQLite with custom journaling\n- Read side: 1000 queries/sec, 3 read projections in shared memory segments\n- Eventual consistency window: 100ms max\n- Handle atomic swap of projection memory for rebuilds\n- Binary configuration format versioning for schema evolution\n- Framework: libevent with custom protocol parser\n\nConstraints:\n- Manual memory management only, no garbage collection\n- Lock-free data structures where possible\n- Shared memory projections must survive process restarts\n- Command handlers must be thread-safe with 4 worker threads\n- Projection rebuild must not block queries\n- Binary format must support forward/backward compatibility\n- Error handling for corrupted journal recovery\n- Memory-mapped I/O for shared segments\n- Zero-copy where possible for performance\n\nDeliverables:\n1. Command processing pipeline with journaling\n2. Projection engine with shared memory management\n3. Query dispatcher with read-your-writes consistency\n4. Schema evolution system with versioned binary format\n5. Integration with libevent for network I/O\n6. Stress test showing 200 cmd/s + 1000 q/s sustained\n\nAssume x86_64 Linux, pthreads, atomic operations. No high-level frameworks."

messages = [
    {"role": "user", "content": prompt},
]

# Process input
inputs = processor.apply_chat_template(
    messages,
    tokenize=True,
    return_dict=True,
    return_tensors="pt",
    add_generation_prompt=True,
    reasoning_strength="high"
).to(model.device)
input_len = inputs["input_ids"].shape[-1]

# Generate output
outputs = model.generate(**inputs)
response = processor.decode(outputs[0][input_len:], skip_special_tokens=False)
print(response)
```


Tachibana-Agent is one of our [Experimental Reasoning Models.](https://huggingface.co/collections/sequelbox/experimental-reasoning-models)

Do as you will.