Add deepseek-ai/DeepSeek-V4-Flash to quant queue
Browse files
deepseek-ai/DeepSeek-V4-Flash_quant_request_False_W4A16_4bit_int4_RTN.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"job_type": "quantization & evaluation", "model": "deepseek-ai/DeepSeek-V4-Flash", "revision": "main", "private": false, "params": 158.07, "architectures": "DeepseekV4ForCausalLM", "num_layers": 43, "input_dtype": "bfloat16", "input_bits": 16, "model_params": 158.07, "model_weight_gb": 186.19, "quant_scheme": "INT4 (W4A16)", "quant_precision": "4bit", "quant_weight_dtype": "int4", "quant_bits": 4, "quant_memory_gb": 12.99, "quant_gpu_type": "NVIDIA GeForce RTX 4090", "quant_gpu_nums": 1, "eval_memory_gb": 173.88, "eval_gpu_type": "NVIDIA H200", "eval_gpu_nums": 2, "quant_model_size_gb": 79.03, "hardware": "NVIDIA H200", "gpu_nums": 2, "method": "RTN", "script": "auto_quant", "status": "Pending", "submitted_time": "2026-08-03T12:14:31Z", "submitted_by": "wenjiao", "submitted_orgs": ["Intel"], "model_type": "quantization", "job_id": -1, "job_start_time": null, "likes": 1980, "license": "mit", "still_on_hub": true, "tags": ["moe"], "downloads": 0, "created_at": ""}
|