Instructions to use eewer/qwen3.5-35b-a3b-projectbro-rl-65k-iter99 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use eewer/qwen3.5-35b-a3b-projectbro-rl-65k-iter99 with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="eewer/qwen3.5-35b-a3b-projectbro-rl-65k-iter99")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("eewer/qwen3.5-35b-a3b-projectbro-rl-65k-iter99")
model = AutoModelForCausalLM.from_pretrained("eewer/qwen3.5-35b-a3b-projectbro-rl-65k-iter99", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use eewer/qwen3.5-35b-a3b-projectbro-rl-65k-iter99 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "eewer/qwen3.5-35b-a3b-projectbro-rl-65k-iter99"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "eewer/qwen3.5-35b-a3b-projectbro-rl-65k-iter99",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/eewer/qwen3.5-35b-a3b-projectbro-rl-65k-iter99

SGLang

How to use eewer/qwen3.5-35b-a3b-projectbro-rl-65k-iter99 with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "eewer/qwen3.5-35b-a3b-projectbro-rl-65k-iter99" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "eewer/qwen3.5-35b-a3b-projectbro-rl-65k-iter99",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "eewer/qwen3.5-35b-a3b-projectbro-rl-65k-iter99" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "eewer/qwen3.5-35b-a3b-projectbro-rl-65k-iter99",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use eewer/qwen3.5-35b-a3b-projectbro-rl-65k-iter99 with Docker Model Runner:
```
docker model run hf.co/eewer/qwen3.5-35b-a3b-projectbro-rl-65k-iter99
```

Qwen3.5-35B-A3B terminal_swe_general agent — step 7,176

This is the step-7,176 final checkpoint of the terminal_swe_general agent run. It is a supervised fine-tune of Qwen/Qwen3.5-35B-A3B-Base for agentic text generation. General agent SFT balanced across Terminus-2, other terminal/SWE harnesses, and BFCL-/Tau2-like agentic tasks.

This repository contains BF16 Hugging Face safetensors exported from the finalized Megatron distributed checkpoint. It is not quantized.

Checkpoint identity

Field	Value
Hugging Face repository	`eewer/qwen3.5-35b-a3b-base-terminal_swe_general-agent-ablation-step7176`
Run name	`terminal_swe_general`
Training iteration	7,176
Planned iterations / epoch	7,176
Position in planned epoch	100.00%
Base model	`Qwen/Qwen3.5-35B-A3B-Base`
Source checkpoint	`/KRAFTON/WORKSPACE/wbl-workspace/posttraining-2606/areal_runs/qwen35_b300/sft/mixed-a-packed-tools-v2-tp1-ep8-cp1-dp8-mbs1-gbs32-lr1e5/checkpoints/iter_0007176`
Export format	BF16 Hugging Face safetensors
Context length used for SFT	32,768 tokens

Training configuration

Hardware: one node with 8 NVIDIA B300 GPUs.
Framework: Megatron-Bridge / Megatron-Core in the NVIDIA NeMo 26.06 environment.
Parallelism: TP=1, PP=1, CP=1, EP=8, DP=8.
Micro batch size: 1 per data-parallel rank.
Global batch size: 32.
Activation recomputation: full, uniform recomputation.
MoE router fusion: enabled; shared-expert overlap, gradient-reduce overlap, and parameter-gather overlap disabled.
Precision: BF16 training/model tensors with the recipe's precision-aware optimizer state; exported weights are BF16.
Learning-rate schedule: cosine decay from a peak of 1e-5 to 1e-6.
Warmup: 158 optimizer steps for every run.
Checkpoint interval: 500 optimizer steps.
Sequence packing: Yes; 229,607 packs, 96.9786% packing efficiency.
Objective: assistant-only causal language modeling. System, user, and tool messages are context rather than prediction targets.

Training data

The materialized source mixture has 505,938 rows; 505,928 rows enter training after the 32K boundary check. The complete training representation contains 7,296,442,884 effective/rendered tokens.

Source dataset	Selected rows
`nvidia/Nemotron-Terminal-Corpus`	79,201
`open-thoughts/OpenThoughts-Agent-SFT-100K`	82,047
`nvidia/Open-SWE-Traces`	35,357
`nvidia/Nemotron-SFT-SWE-v3`	55,558
`allenai/tmax-sft`	655
`allenai/tmax-sft-big`	84,474
`nvidia/Nemotron-SFT-Agentic-v2`	168,646

The source data is deduplicated and deterministically shuffled. Its three 168,646-row buckets cover Terminus-2, other terminal/SWE harnesses, and general agentic tasks. The general bucket is divided between BFCL-like tool calling and Tau2-like interaction. Every assistant turn has non-empty reasoning. Sequences are packed to 32,768 tokens and all assistant outputs are supervised without per-turn loss masks. Ten overlength boundary cases were excluded. Missing Nemotron-SWE tool schemas were reconstructed only from byte-exact system-prompt matches to Open-SWE; unsafe schema matches were rejected.

The tokenizer is from Qwen3.5, with a Qwen3.6 thinking-preserving chat template. The template keeps reasoning content in assistant messages and preserves tool declarations when the source supplies a validated schema.

Intended use

This checkpoint is intended for research on terminal, software-engineering, tool-use, and interactive agents. Use the included chat template and supply tool schemas expected by the target harness. It can be served with Transformers-compatible runtimes that support the Qwen3.5 MoE architecture.

This is a training checkpoint, not a polished instruction model. Compare checkpoints using held-out loss and task-level agent evaluations before selecting one for deployment.

Limitations and safety

No complete benchmark suite is claimed in this model card.
Agent trajectories can produce destructive shell commands, modify files, call tools, or expose secrets. Run the model in an isolated environment with scoped credentials.
Tool names and schemas vary across source harnesses. A caller must provide the schema appropriate for its own environment; do not assume a generated call is valid or safe.
Dataset filtering and deduplication reduce, but cannot guarantee removal of, benchmark contamination, incorrect reasoning, insecure code, or teacher-model artifacts.
The model can hallucinate successful tool execution and should not be trusted without checking actual environment observations.
This checkpoint inherits the base model's limitations and license.

Reproducibility notes

The B300 SFT recipes use the same execution topology, optimizer family, 158-step warmup, and checkpoint cadence. The checkpoint-specific learning-rate range and packing mode are reported above. The terminal_swe and terminal_swe_general runs differ from the terminal runs primarily in their source mixtures. The mixed datasets were validated for unique conversation hashes, deterministic shuffle order, non-empty reasoning in every assistant turn, maximum rendered length, and tool-call/schema consistency.

Downloads last month: 21

Safetensors

Model size

35B params

Tensor type

BF16

Model tree for eewer/qwen3.5-35b-a3b-projectbro-rl-65k-iter99

Base model

Qwen/Qwen3.5-35B-A3B-Base

Finetuned

(77)

this model

eewer
/

qwen3.5-35b-a3b-projectbro-rl-65k-iter99