Nanbeige4.3-33B-FP8 / config.json
GetCortex27's picture
FP8_DYNAMIC quant of Nanbeige4.3-33B (llm-compressor), with fidelity + throughput card
324f2c1 verified
Raw History Blame Contribute Delete
1.88 kB
{
"architectures": [
"LlamaForCausalLM"
],
"attention_bias": false,
"attention_dropout": 0.0,
"bos_token_id": 166100,
"embd_pdrop": 0.0,
"eos_token_id": 166101,
"ffn_hidden_size": 22016,
"head_dim": 128,
"hidden_act": "silu",
"hidden_size": 7168,
"initializer_range": 0.02,
"intermediate_size": 22016,
"kv_channels": 128,
"max_position_embeddings": 524288,
"mlp_bias": false,
"model_type": "llama",
"num_attention_heads": 56,
"num_hidden_layers": 52,
"num_key_value_heads": 8,
"num_query_groups": 8,
"pad_token_id": 0,
"pretraining_tp": 1,
"quantization_config": {
"config_groups": {
"group_0": {
"format": "float-quantized",
"input_activations": {
"dynamic": true,
"group_size": null,
"num_bits": 8,
"strategy": "token",
"symmetric": true,
"type": "float"
},
"output_activations": null,
"targets": [
"Linear"
],
"weights": {
"dynamic": false,
"group_size": null,
"num_bits": 8,
"observer": "memoryless_minmax",
"strategy": "channel",
"symmetric": true,
"type": "float"
}
}
},
"format": "float-quantized",
"global_compression_ratio": null,
"ignore": [
"lm_head"
],
"kv_cache_scheme": null,
"moe_sparsity_config": null,
"quant_method": "compressed-tensors",
"quantization_status": "compressed",
"sparsity_config": {},
"transform_config": {},
"version": "0.20.1"
},
"resid_pdrop": 0.0,
"rms_norm_eps": 1e-05,
"rope_scaling": null,
"rope_theta": 120000000,
"rotary_base": 120000000,
"seq_length": 524288,
"tie_word_embeddings": false,
"torch_dtype": "bfloat16",
"transformers_version": "4.51.1",
"use_cache": true,
"vocab_size": 166912
}