Qwen3.8-27B-opencode-adapted-bf16

Qwen3.8-27B-opencode-adapted-bf16 is a standalone BF16 checkpoint created by merging the harness-hub/Qwen3.8-27B-opencode-adapted LoRA adapter into Qwen/Qwen3.8-27B, enabling deployment on any inference engine without LoRA support. The adapter (r = 32, alpha = 64, targeting the q, k, v, o, gate, up, and down projections) was trained with AsyncGRPO reinforcement learning in Harbor/OpenEnv environments to serve as the backbone for the OpenCode coding agent on software-engineering and terminal tasks, with thinking mode turned off as in training. The merge applies the standard W + (alpha/r)Β·B@A update per tensor in fp32 before casting back to BF16, preserving the base shard layout, tensor names, and all MTP tensors. According to the adapter's model card, under OpenCode it raises SWE-bench Pro (hard 51) from 35.3% to 51.0% and Terminal-Bench 2.1 from 46.4% to 52.4%, while requiring a quarter to half as many turns and generating 6–8Γ— fewer tokens per task. These figures were measured using the adapter on the BF16 base model and have not been re-evaluated on this merged checkpoint.

Model Files

File Name File Type File Size File Link Description
chat_template.jinja Jinja Template 9 kB Link Chat template for formatting conversations and prompts.
config.json JSON 4.31 kB Link Core model architecture and configuration parameters.
generation_config.json JSON 202 Bytes Link Default text generation parameters (e.g., temperature, top_p).
merges.txt Text 3.35 MB Link Byte-Pair Encoding (BPE) merge rules for the tokenizer.
model-00001-of-00018.safetensors Safetensors 3.97 GB Link Model weight shard 1/18. Required to load the model.
model-00002-of-00018.safetensors Safetensors 3.04 GB Link Model weight shard 2/18. Required to load the model.
model-00003-of-00018.safetensors Safetensors 2.54 GB Link Model weight shard 3/18. Required to load the model.
model-00004-of-00018.safetensors Safetensors 3.99 GB Link Model weight shard 4/18. Required to load the model.
model-00005-of-00018.safetensors Safetensors 2.1 GB Link Model weight shard 5/18. Required to load the model.
model-00006-of-00018.safetensors Safetensors 3.98 GB Link Model weight shard 6/18. Required to load the model.
model-00007-of-00018.safetensors Safetensors 2.11 GB Link Model weight shard 7/18. Required to load the model.
model-00008-of-00018.safetensors Safetensors 3.98 GB Link Model weight shard 8/18. Required to load the model.
model-00009-of-00018.safetensors Safetensors 2.11 GB Link Model weight shard 9/18. Required to load the model.
model-00010-of-00018.safetensors Safetensors 3.98 GB Link Model weight shard 10/18. Required to load the model.
model-00011-of-00018.safetensors Safetensors 2.11 GB Link Model weight shard 11/18. Required to load the model.
model-00012-of-00018.safetensors Safetensors 3.98 GB Link Model weight shard 12/18. Required to load the model.
model-00013-of-00018.safetensors Safetensors 2.11 GB Link Model weight shard 13/18. Required to load the model.
model-00014-of-00018.safetensors Safetensors 3.98 GB Link Model weight shard 14/18. Required to load the model.
model-00015-of-00018.safetensors Safetensors 2.11 GB Link Model weight shard 15/18. Required to load the model.
model-00016-of-00018.safetensors Safetensors 3.98 GB Link Model weight shard 16/18. Required to load the model.
model-00017-of-00018.safetensors Safetensors 2.11 GB Link Model weight shard 17/18. Required to load the model.
model-00018-of-00018.safetensors Safetensors 3.39 GB Link Model weight shard 18/18. Required to load the model.
model.safetensors.index.json JSON 112 kB Link Index file mapping model weights to their respective shards.
preprocessor_config.json JSON 390 Bytes Link Configuration for preprocessing vision/image inputs.
tokenizer.json JSON 12.8 MB Link Complete tokenizer vocabulary and encoding rules.
tokenizer_config.json JSON 9.98 kB Link Tokenizer-specific configuration parameters.
video_preprocessor_config.json JSON 385 Bytes Link Configuration for preprocessing video inputs.
vocab.json JSON 6.72 MB Link Tokenizer vocabulary mapping tokens to IDs.

License and Attribution

This model is based on and/or incorporates the following open-source projects and models:

This model is released under the Apache License 2.0.

Downloads last month
25
Safetensors
Model size
28B params
Tensor type
BF16
Β·
Inference Providers NEW
This model isn't deployed by any Inference Provider. πŸ™‹ Ask for provider support

Model tree for prithivMLmods/Qwen3.8-27B-opencode-adapted-bf16

Base model

Qwen/Qwen3.8-27B
Adapter
(2)
this model
Adapters
2 models

Datasets used to train prithivMLmods/Qwen3.8-27B-opencode-adapted-bf16

Collections including prithivMLmods/Qwen3.8-27B-opencode-adapted-bf16