QUASAR-QAT/gemma-4-12B-it-QUASAR-Q4_0-GGUF

zetlyn/models-gguf quantisation gguf QUASAR-QAT/gemma-4-12B-it-QUASAR-Q4_0-GGUF hf google/gemma-4-12B-it known 2026-09-11

https://huggingface.co/QUASAR-QAT/gemma-4-12B-it-QUASAR-Q4_0-GGUF

Properties

author	QUASAR-QAT receipt Source GGUF quantisations Its words `QUASAR-QAT` Read by `field:author` Said since 2026-09-29 09:50 UTC Last answered 2026-10-04 12:16 UTC Original open at the source What the source handed over { "_id": "6aa34996b15c5e5397190836", "author": "QUASAR-QAT", "cardData": { "base_model": "google/gemma-4-12B-it", "base_model_relation": "quantized", "language": [ "en", "multilingual" ], "license": "apache-2.0", "license_link": "https://huggingface.co/google/gemma-4-12B-it", "model-index": [ { "name": "gemma-4-12B-it-QUASAR-Q4_0-GGUF", "results": [ { "dataset": { "name": "Held-out chat prompts (fidelity to BF16 through llama.cpp)", "type": "QUASAR-QAT/gemma-4-12B-it-QUASAR-Q4_0-GGUF" }, "metrics": [ { "name": "KL(BF16 ‖ model), llama.cpp Q4_0 (lower is better)", "type": "kl_divergence", "value": 0.048, "verified": false }, { "name": "top-1 agreement with BF16", "type": "accuracy", "value": 94.3, "verified": false } ], "source": { "name": "EVAL.md (our measurements, same harness for every row)", "url": "https://huggingface.co/QUASAR-QAT/gemma-4-12B-it-QUASAR-Q4_0-GGUF/blob/main/EVAL.md" }, "task": { "name": "text generation", "type": "text-generation" } } ] } ], "pipeline_tag": "image-text-to-text", "quantized_by": "QUASAR-QAT", "tags": [ "quasar", "native-qat", "qat", "quantization-aware-training", "int4", "gemma-4", "gemma4", "gemma", "google", "conversational", "gguf", "q4_0", "llama.cpp", "ollama", "lm-studio", "12b", "gemma-4-12b", "gemma-4-12b-it", "quantized", "4-bit precision", "weight-only", "multimodal", "reasoning", "thinking", "long-context", "function-calling", "tool-calling", "agentic", "vision", "vlm", "audio", "text-generation" ] }, "createdAt": "2026-09-11T00:21:42.000Z", "downloads": 1316, "gated": false, "id": "QUASAR-QAT/gemma-4-12B-it-QUASAR-Q4_0-GGUF", "lastModified": "2026-09-16T19:19:32.000Z", "likes": 4, "modelId": "QUASAR-QAT/gemma-4-12B-it-QUASAR-Q4_0-GGUF", "pipeline_tag": "image-text-to-text", "private": false, "sha": "0870350340956c1102a848f898c62d365fb27cfd", "siblings": [ { "rfilename": ".gitattributes" }, { "rfilename": "EVAL.md" }, { "rfilename": "Modelfile" }, { "rfilename": "README.md" }, { "rfilename": "SHA256SUMS" }, { "rfilename": "gemma-4-12B-it-QUASAR-Q4_0.gguf" }, { "rfilename": "kl_12b.svg" }, { "rfilename": "mmproj-gemma-4-12B-it-BF16.gguf" }, { "rfilename": "native_q4_0_receipt.json" }, { "rfilename": "scripts/export_gemma4_gguf_q4_0.py" }, { "rfilename": "scripts/harvest_12b_all.py" }, { "rfilename": "scripts/harvest_12b_chat.py" }, { "rfilename": "scripts/paired_bootstrap.py" } ], "tags": [ "gguf", "quasar", "native-qat", "qat", "quantization-aware-training", "int4", "gemma-4", "gemma4", "gemma", "google", "conversational", "q4_0", "llama.cpp", "ollama", "lm-studio", "12b", "gemma-4-12b", "gemma-4-12b-it", "quantized", "4-bit precision", "weight-only", "multimodal", "reasoning", "thinking", "long-context", "function-calling", "tool-calling", "agentic", "vision", "vlm", "audio", "text-generation", "image-text-to-text", "en", "multilingual", "arxiv:2608.13966", "base_model:google/gemma-4-12B-it", "base_model:quantized:google/gemma-4-12B-it", "license:apache-2.0", "model-index", "endpoints_compatible", "region:us" ] }
base	google/gemma-4-12B-it receipt Source GGUF quantisations Its words `google/gemma-4-12B-it` Read by `field:cardData.base_model[]` Said since 2026-09-29 09:50 UTC Last answered 2026-10-04 12:16 UTC Original open at the source What the source handed over { "_id": "6aa34996b15c5e5397190836", "author": "QUASAR-QAT", "cardData": { "base_model": "google/gemma-4-12B-it", "base_model_relation": "quantized", "language": [ "en", "multilingual" ], "license": "apache-2.0", "license_link": "https://huggingface.co/google/gemma-4-12B-it", "model-index": [ { "name": "gemma-4-12B-it-QUASAR-Q4_0-GGUF", "results": [ { "dataset": { "name": "Held-out chat prompts (fidelity to BF16 through llama.cpp)", "type": "QUASAR-QAT/gemma-4-12B-it-QUASAR-Q4_0-GGUF" }, "metrics": [ { "name": "KL(BF16 ‖ model), llama.cpp Q4_0 (lower is better)", "type": "kl_divergence", "value": 0.048, "verified": false }, { "name": "top-1 agreement with BF16", "type": "accuracy", "value": 94.3, "verified": false } ], "source": { "name": "EVAL.md (our measurements, same harness for every row)", "url": "https://huggingface.co/QUASAR-QAT/gemma-4-12B-it-QUASAR-Q4_0-GGUF/blob/main/EVAL.md" }, "task": { "name": "text generation", "type": "text-generation" } } ] } ], "pipeline_tag": "image-text-to-text", "quantized_by": "QUASAR-QAT", "tags": [ "quasar", "native-qat", "qat", "quantization-aware-training", "int4", "gemma-4", "gemma4", "gemma", "google", "conversational", "gguf", "q4_0", "llama.cpp", "ollama", "lm-studio", "12b", "gemma-4-12b", "gemma-4-12b-it", "quantized", "4-bit precision", "weight-only", "multimodal", "reasoning", "thinking", "long-context", "function-calling", "tool-calling", "agentic", "vision", "vlm", "audio", "text-generation" ] }, "createdAt": "2026-09-11T00:21:42.000Z", "downloads": 1316, "gated": false, "id": "QUASAR-QAT/gemma-4-12B-it-QUASAR-Q4_0-GGUF", "lastModified": "2026-09-16T19:19:32.000Z", "likes": 4, "modelId": "QUASAR-QAT/gemma-4-12B-it-QUASAR-Q4_0-GGUF", "pipeline_tag": "image-text-to-text", "private": false, "sha": "0870350340956c1102a848f898c62d365fb27cfd", "siblings": [ { "rfilename": ".gitattributes" }, { "rfilename": "EVAL.md" }, { "rfilename": "Modelfile" }, { "rfilename": "README.md" }, { "rfilename": "SHA256SUMS" }, { "rfilename": "gemma-4-12B-it-QUASAR-Q4_0.gguf" }, { "rfilename": "kl_12b.svg" }, { "rfilename": "mmproj-gemma-4-12B-it-BF16.gguf" }, { "rfilename": "native_q4_0_receipt.json" }, { "rfilename": "scripts/export_gemma4_gguf_q4_0.py" }, { "rfilename": "scripts/harvest_12b_all.py" }, { "rfilename": "scripts/harvest_12b_chat.py" }, { "rfilename": "scripts/paired_bootstrap.py" } ], "tags": [ "gguf", "quasar", "native-qat", "qat", "quantization-aware-training", "int4", "gemma-4", "gemma4", "gemma", "google", "conversational", "q4_0", "llama.cpp", "ollama", "lm-studio", "12b", "gemma-4-12b", "gemma-4-12b-it", "quantized", "4-bit precision", "weight-only", "multimodal", "reasoning", "thinking", "long-context", "function-calling", "tool-calling", "agentic", "vision", "vlm", "audio", "text-generation", "image-text-to-text", "en", "multilingual", "arxiv:2608.13966", "base_model:google/gemma-4-12B-it", "base_model:quantized:google/gemma-4-12B-it", "license:apache-2.0", "model-index", "endpoints_compatible", "region:us" ] }
downloads	1316 receipt Source GGUF quantisations Its words `1316` Read by `field:downloads` Said since 2026-09-29 09:50 UTC Last answered 2026-10-04 12:16 UTC Original open at the source What the source handed over { "_id": "6aa34996b15c5e5397190836", "author": "QUASAR-QAT", "cardData": { "base_model": "google/gemma-4-12B-it", "base_model_relation": "quantized", "language": [ "en", "multilingual" ], "license": "apache-2.0", "license_link": "https://huggingface.co/google/gemma-4-12B-it", "model-index": [ { "name": "gemma-4-12B-it-QUASAR-Q4_0-GGUF", "results": [ { "dataset": { "name": "Held-out chat prompts (fidelity to BF16 through llama.cpp)", "type": "QUASAR-QAT/gemma-4-12B-it-QUASAR-Q4_0-GGUF" }, "metrics": [ { "name": "KL(BF16 ‖ model), llama.cpp Q4_0 (lower is better)", "type": "kl_divergence", "value": 0.048, "verified": false }, { "name": "top-1 agreement with BF16", "type": "accuracy", "value": 94.3, "verified": false } ], "source": { "name": "EVAL.md (our measurements, same harness for every row)", "url": "https://huggingface.co/QUASAR-QAT/gemma-4-12B-it-QUASAR-Q4_0-GGUF/blob/main/EVAL.md" }, "task": { "name": "text generation", "type": "text-generation" } } ] } ], "pipeline_tag": "image-text-to-text", "quantized_by": "QUASAR-QAT", "tags": [ "quasar", "native-qat", "qat", "quantization-aware-training", "int4", "gemma-4", "gemma4", "gemma", "google", "conversational", "gguf", "q4_0", "llama.cpp", "ollama", "lm-studio", "12b", "gemma-4-12b", "gemma-4-12b-it", "quantized", "4-bit precision", "weight-only", "multimodal", "reasoning", "thinking", "long-context", "function-calling", "tool-calling", "agentic", "vision", "vlm", "audio", "text-generation" ] }, "createdAt": "2026-09-11T00:21:42.000Z", "downloads": 1316, "gated": false, "id": "QUASAR-QAT/gemma-4-12B-it-QUASAR-Q4_0-GGUF", "lastModified": "2026-09-16T19:19:32.000Z", "likes": 4, "modelId": "QUASAR-QAT/gemma-4-12B-it-QUASAR-Q4_0-GGUF", "pipeline_tag": "image-text-to-text", "private": false, "sha": "0870350340956c1102a848f898c62d365fb27cfd", "siblings": [ { "rfilename": ".gitattributes" }, { "rfilename": "EVAL.md" }, { "rfilename": "Modelfile" }, { "rfilename": "README.md" }, { "rfilename": "SHA256SUMS" }, { "rfilename": "gemma-4-12B-it-QUASAR-Q4_0.gguf" }, { "rfilename": "kl_12b.svg" }, { "rfilename": "mmproj-gemma-4-12B-it-BF16.gguf" }, { "rfilename": "native_q4_0_receipt.json" }, { "rfilename": "scripts/export_gemma4_gguf_q4_0.py" }, { "rfilename": "scripts/harvest_12b_all.py" }, { "rfilename": "scripts/harvest_12b_chat.py" }, { "rfilename": "scripts/paired_bootstrap.py" } ], "tags": [ "gguf", "quasar", "native-qat", "qat", "quantization-aware-training", "int4", "gemma-4", "gemma4", "gemma", "google", "conversational", "q4_0", "llama.cpp", "ollama", "lm-studio", "12b", "gemma-4-12b", "gemma-4-12b-it", "quantized", "4-bit precision", "weight-only", "multimodal", "reasoning", "thinking", "long-context", "function-calling", "tool-calling", "agentic", "vision", "vlm", "audio", "text-generation", "image-text-to-text", "en", "multilingual", "arxiv:2608.13966", "base_model:google/gemma-4-12B-it", "base_model:quantized:google/gemma-4-12B-it", "license:apache-2.0", "model-index", "endpoints_compatible", "region:us" ] }
licence	apache-2.0 receipt Source GGUF quantisations Its words `apache-2.0` Read by `field:cardData.license` Said since 2026-09-29 09:50 UTC Last answered 2026-10-04 12:16 UTC Original open at the source What the source handed over { "_id": "6aa34996b15c5e5397190836", "author": "QUASAR-QAT", "cardData": { "base_model": "google/gemma-4-12B-it", "base_model_relation": "quantized", "language": [ "en", "multilingual" ], "license": "apache-2.0", "license_link": "https://huggingface.co/google/gemma-4-12B-it", "model-index": [ { "name": "gemma-4-12B-it-QUASAR-Q4_0-GGUF", "results": [ { "dataset": { "name": "Held-out chat prompts (fidelity to BF16 through llama.cpp)", "type": "QUASAR-QAT/gemma-4-12B-it-QUASAR-Q4_0-GGUF" }, "metrics": [ { "name": "KL(BF16 ‖ model), llama.cpp Q4_0 (lower is better)", "type": "kl_divergence", "value": 0.048, "verified": false }, { "name": "top-1 agreement with BF16", "type": "accuracy", "value": 94.3, "verified": false } ], "source": { "name": "EVAL.md (our measurements, same harness for every row)", "url": "https://huggingface.co/QUASAR-QAT/gemma-4-12B-it-QUASAR-Q4_0-GGUF/blob/main/EVAL.md" }, "task": { "name": "text generation", "type": "text-generation" } } ] } ], "pipeline_tag": "image-text-to-text", "quantized_by": "QUASAR-QAT", "tags": [ "quasar", "native-qat", "qat", "quantization-aware-training", "int4", "gemma-4", "gemma4", "gemma", "google", "conversational", "gguf", "q4_0", "llama.cpp", "ollama", "lm-studio", "12b", "gemma-4-12b", "gemma-4-12b-it", "quantized", "4-bit precision", "weight-only", "multimodal", "reasoning", "thinking", "long-context", "function-calling", "tool-calling", "agentic", "vision", "vlm", "audio", "text-generation" ] }, "createdAt": "2026-09-11T00:21:42.000Z", "downloads": 1316, "gated": false, "id": "QUASAR-QAT/gemma-4-12B-it-QUASAR-Q4_0-GGUF", "lastModified": "2026-09-16T19:19:32.000Z", "likes": 4, "modelId": "QUASAR-QAT/gemma-4-12B-it-QUASAR-Q4_0-GGUF", "pipeline_tag": "image-text-to-text", "private": false, "sha": "0870350340956c1102a848f898c62d365fb27cfd", "siblings": [ { "rfilename": ".gitattributes" }, { "rfilename": "EVAL.md" }, { "rfilename": "Modelfile" }, { "rfilename": "README.md" }, { "rfilename": "SHA256SUMS" }, { "rfilename": "gemma-4-12B-it-QUASAR-Q4_0.gguf" }, { "rfilename": "kl_12b.svg" }, { "rfilename": "mmproj-gemma-4-12B-it-BF16.gguf" }, { "rfilename": "native_q4_0_receipt.json" }, { "rfilename": "scripts/export_gemma4_gguf_q4_0.py" }, { "rfilename": "scripts/harvest_12b_all.py" }, { "rfilename": "scripts/harvest_12b_chat.py" }, { "rfilename": "scripts/paired_bootstrap.py" } ], "tags": [ "gguf", "quasar", "native-qat", "qat", "quantization-aware-training", "int4", "gemma-4", "gemma4", "gemma", "google", "conversational", "q4_0", "llama.cpp", "ollama", "lm-studio", "12b", "gemma-4-12b", "gemma-4-12b-it", "quantized", "4-bit precision", "weight-only", "multimodal", "reasoning", "thinking", "long-context", "function-calling", "tool-calling", "agentic", "vision", "vlm", "audio", "text-generation", "image-text-to-text", "en", "multilingual", "arxiv:2608.13966", "base_model:google/gemma-4-12B-it", "base_model:quantized:google/gemma-4-12B-it", "license:apache-2.0", "model-index", "endpoints_compatible", "region:us" ] }
likes	4 receipt Source GGUF quantisations Its words `4` Read by `field:likes` Said since 2026-09-29 09:50 UTC Last answered 2026-10-04 12:16 UTC Original open at the source What the source handed over { "_id": "6aa34996b15c5e5397190836", "author": "QUASAR-QAT", "cardData": { "base_model": "google/gemma-4-12B-it", "base_model_relation": "quantized", "language": [ "en", "multilingual" ], "license": "apache-2.0", "license_link": "https://huggingface.co/google/gemma-4-12B-it", "model-index": [ { "name": "gemma-4-12B-it-QUASAR-Q4_0-GGUF", "results": [ { "dataset": { "name": "Held-out chat prompts (fidelity to BF16 through llama.cpp)", "type": "QUASAR-QAT/gemma-4-12B-it-QUASAR-Q4_0-GGUF" }, "metrics": [ { "name": "KL(BF16 ‖ model), llama.cpp Q4_0 (lower is better)", "type": "kl_divergence", "value": 0.048, "verified": false }, { "name": "top-1 agreement with BF16", "type": "accuracy", "value": 94.3, "verified": false } ], "source": { "name": "EVAL.md (our measurements, same harness for every row)", "url": "https://huggingface.co/QUASAR-QAT/gemma-4-12B-it-QUASAR-Q4_0-GGUF/blob/main/EVAL.md" }, "task": { "name": "text generation", "type": "text-generation" } } ] } ], "pipeline_tag": "image-text-to-text", "quantized_by": "QUASAR-QAT", "tags": [ "quasar", "native-qat", "qat", "quantization-aware-training", "int4", "gemma-4", "gemma4", "gemma", "google", "conversational", "gguf", "q4_0", "llama.cpp", "ollama", "lm-studio", "12b", "gemma-4-12b", "gemma-4-12b-it", "quantized", "4-bit precision", "weight-only", "multimodal", "reasoning", "thinking", "long-context", "function-calling", "tool-calling", "agentic", "vision", "vlm", "audio", "text-generation" ] }, "createdAt": "2026-09-11T00:21:42.000Z", "downloads": 1316, "gated": false, "id": "QUASAR-QAT/gemma-4-12B-it-QUASAR-Q4_0-GGUF", "lastModified": "2026-09-16T19:19:32.000Z", "likes": 4, "modelId": "QUASAR-QAT/gemma-4-12B-it-QUASAR-Q4_0-GGUF", "pipeline_tag": "image-text-to-text", "private": false, "sha": "0870350340956c1102a848f898c62d365fb27cfd", "siblings": [ { "rfilename": ".gitattributes" }, { "rfilename": "EVAL.md" }, { "rfilename": "Modelfile" }, { "rfilename": "README.md" }, { "rfilename": "SHA256SUMS" }, { "rfilename": "gemma-4-12B-it-QUASAR-Q4_0.gguf" }, { "rfilename": "kl_12b.svg" }, { "rfilename": "mmproj-gemma-4-12B-it-BF16.gguf" }, { "rfilename": "native_q4_0_receipt.json" }, { "rfilename": "scripts/export_gemma4_gguf_q4_0.py" }, { "rfilename": "scripts/harvest_12b_all.py" }, { "rfilename": "scripts/harvest_12b_chat.py" }, { "rfilename": "scripts/paired_bootstrap.py" } ], "tags": [ "gguf", "quasar", "native-qat", "qat", "quantization-aware-training", "int4", "gemma-4", "gemma4", "gemma", "google", "conversational", "q4_0", "llama.cpp", "ollama", "lm-studio", "12b", "gemma-4-12b", "gemma-4-12b-it", "quantized", "4-bit precision", "weight-only", "multimodal", "reasoning", "thinking", "long-context", "function-calling", "tool-calling", "agentic", "vision", "vlm", "audio", "text-generation", "image-text-to-text", "en", "multilingual", "arxiv:2608.13966", "base_model:google/gemma-4-12B-it", "base_model:quantized:google/gemma-4-12B-it", "license:apache-2.0", "model-index", "endpoints_compatible", "region:us" ] }

Text

This source's terms allow its title, its values and a link here, not its text. It is at https://huggingface.co/QUASAR-QAT/gemma-4-12B-it-QUASAR-Q4_0-GGUF.