---

license: other

language:

• en

library_name: llama.cpp

pipeline_tag: text-generation

base_model:

• TaimoorSiddiqui/HopCoder-Mini-35B-A3B-VL36
• TaimoorSiddiqui/HopCoder-Mini-35B-A3B-VL36-fullsft

datasets:

• TaimoorSiddiqui/hopcoder-fullsft-mix

tags:

• gguf
• llama.cpp
• qwen3.5
• qwen3.5-moe
• moe
• vision-language
• full-finetuning
• unsloth
• modal
• standard-gguf-quantization

---

HopCoder Mini 35B A3B VL36 Full-SFT GGUF

This repository contains GGUF conversions of TaimoorSiddiqui/HopCoder-Mini-35B-A3B-VL36-fullsft (400-step fine-tune).

Generated from the checkpoint-400 F16 GGUF file without importance matrix calibration. Low-bit 2-bit quants (IQ2_, IQ3_, IQ4_*, Q2_K_S) that require an importance matrix are not included.

Tool Calling And Agent Runtimes

The source tokenizer/chat template is copied into this repo, and the GGUF also embeds a chat template. The current tool-call shape is JSON inside <tool_call> tags:

<tool_call>
{"name":"tool_name","arguments":{"argument_name":"value"}}
</tool_call>

Files

| Quant | Status | Total Size | Files |

| --- | --- | ---: | --- |

| BF16 | uploaded | 1.03 GB | HopCoder-Mini-35B-A3B-VL36-fullsft-BF16.gguf |

| F16-00001-of-00002 | uploaded | 17.80 GB | HopCoder-Mini-35B-A3B-VL36-fullsft-F16-00001-of-00002.gguf |

| F16-00002-of-00002 | uploaded | | HopCoder-Mini-35B-A3B-VL36-fullsft-F16-00002-of-00002.gguf |

| F16 | uploaded | 69.38 GB | HopCoder-Mini-35B-A3B-VL36-fullsft-F16.gguf |

| Q2_K | uploaded | 0.01 GB | HopCoder-Mini-35B-A3B-VL36-fullsft-Q2_K.gguf |

| Q3_K_L | uploaded | 18.12 GB | HopCoder-Mini-35B-A3B-VL36-fullsft-Q3_K_L.gguf |

| Q3_K_M | uploaded | 16.76 GB | HopCoder-Mini-35B-A3B-VL36-fullsft-Q3_K_M.gguf |

| Q3_K_S | uploaded | 15.18 GB | HopCoder-Mini-35B-A3B-VL36-fullsft-Q3_K_S.gguf |

| Q4_0 | uploaded | 19.72 GB | HopCoder-Mini-35B-A3B-VL36-fullsft-Q4_0.gguf |

| Q4_1 | uploaded | 21.85 GB | HopCoder-Mini-35B-A3B-VL36-fullsft-Q4_1.gguf |

| Q4_K_M | uploaded | 21.17 GB | HopCoder-Mini-35B-A3B-VL36-fullsft-Q4_K_M.gguf |

| Q4_K_S | uploaded | 19.89 GB | HopCoder-Mini-35B-A3B-VL36-fullsft-Q4_K_S.gguf |

| Q5_0 | uploaded | 23.98 GB | HopCoder-Mini-35B-A3B-VL36-fullsft-Q5_0.gguf |

| Q5_1 | uploaded | 26.11 GB | HopCoder-Mini-35B-A3B-VL36-fullsft-Q5_1.gguf |

| Q5_K_M | uploaded | 24.73 GB | HopCoder-Mini-35B-A3B-VL36-fullsft-Q5_K_M.gguf |

| Q5_K_S | uploaded | 23.98 GB | HopCoder-Mini-35B-A3B-VL36-fullsft-Q5_K_S.gguf |

| Q6_K | uploaded | 28.51 GB | HopCoder-Mini-35B-A3B-VL36-fullsft-Q6_K.gguf |

| Q8_0 | uploaded | 36.90 GB | HopCoder-Mini-35B-A3B-VL36-fullsft-Q8_0.gguf |

| TQ2_0 | uploaded | 9.47 GB | HopCoder-Mini-35B-A3B-VL36-fullsft-TQ2_0.gguf |

Source

• Full-SFT repo: TaimoorSiddiqui/HopCoder-Mini-35B-A3B-VL36-fullsft
• GGUF repo: TaimoorSiddiqui/HopCoder-Mini-35B-A3B-VL36-fullsft-GGUF
• Training dataset: TaimoorSiddiqui/hopcoder-fullsft-mix

License

The training mix includes apache-2.0, agpl-3.0, mit, and other sources. This model card uses license: other; do not treat this derivative as permissive-only.