AlekseyCalvin's picture
Upload NF4 quantized model and tokenizer
e93659c verified
Raw
History Blame Contribute Delete
3.67 kB
{
"architectures": [
"BltForCausalLM"
],
"cross_attn_k": 4,
"decoder_config": {
"cross_attn_all_layers": true,
"cross_attn_k": 4,
"dropout": 0.0,
"dtype": "float16",
"head_dim": 64,
"hidden_act": "silu",
"hidden_size": 1280,
"hidden_size_global": 4096,
"initializer_range": 0.02,
"intermediate_size": 3584,
"max_position_embeddings": 24576,
"model_type": "blt_local_decoder",
"norm_eps": 1e-05,
"num_attention_heads": 20,
"num_hidden_layers": 6,
"num_key_value_heads": 20,
"rms_norm_eps": 1e-05,
"rope_scaling": {
"rope_type": "default"
},
"rope_theta": 500000.0,
"vocab_size": 260
},
"dtype": "float16",
"encoder_config": {
"cross_attn_all_layers": false,
"cross_attn_k": 4,
"dropout": 0.0,
"dtype": "float16",
"head_dim": 64,
"hidden_act": "silu",
"hidden_size": 1280,
"hidden_size_global": 4096,
"initializer_range": 0.02,
"intermediate_size": 3584,
"max_position_embeddings": 24576,
"model_type": "blt_local_encoder",
"norm_eps": 1e-05,
"num_attention_heads": 20,
"num_hidden_layers": 1,
"num_key_value_heads": 20,
"pm_size": 0,
"rms_norm_eps": 1e-05,
"rope_scaling": {
"rope_type": "default"
},
"rope_theta": 500000.0,
"vocab_size": 260
},
"encoder_hash_byte_group_nb_functions": 1,
"encoder_hash_byte_group_size": [
3,
4,
5,
6,
7,
8
],
"encoder_hash_byte_group_vocab": 500002,
"global_config": {
"dropout": 0.0,
"dtype": "float16",
"encoder_cross_output_size": 5120,
"head_dim": 128,
"hidden_act": "silu",
"hidden_size": 4096,
"initializer_range": 0.02,
"intermediate_size": 11008,
"max_position_embeddings": 4096,
"model_type": "blt_global_transformer",
"norm_eps": 1e-05,
"num_attention_heads": 32,
"num_hidden_layers": 32,
"num_key_value_heads": 32,
"rms_norm_eps": 1e-05,
"rope_scaling": {
"rope_type": "default"
},
"rope_theta": 500000.0
},
"initializer_range": 0.02,
"max_patch_length": null,
"max_position_embeddings": 4096,
"model_type": "blt",
"monotonicity": false,
"patch_in_forward": true,
"patch_size": 4,
"patcher_config": {
"attn_bias_type": "local_block_causal",
"attn_impl": "xformers",
"dropout": 0.0,
"dtype": "float16",
"head_dim": 64,
"hidden_act": "silu",
"hidden_size": 768,
"initializer_range": 0.02,
"intermediate_size": 2048,
"max_position_embeddings": 8192,
"model_type": "blt_patcher",
"norm_eps": 1e-05,
"num_attention_heads": 12,
"num_hidden_layers": 14,
"num_key_value_heads": 12,
"rms_norm_eps": 1e-05,
"rope_scaling": null,
"rope_theta": 10000.0,
"vocab_size": 260
},
"patching_batch_size": 1,
"patching_device": "cuda",
"patching_mode": "entropy",
"patching_threshold": 1.335442066192627,
"patching_threshold_add": null,
"pm_size": 0,
"quantization_config": {
"_load_in_4bit": true,
"_load_in_8bit": false,
"bnb_4bit_compute_dtype": "bfloat16",
"bnb_4bit_quant_storage": "uint8",
"bnb_4bit_quant_type": "nf4",
"bnb_4bit_use_double_quant": false,
"llm_int8_enable_fp32_cpu_offload": false,
"llm_int8_has_fp16_weight": false,
"llm_int8_skip_modules": null,
"llm_int8_threshold": 6.0,
"load_in_4bit": true,
"load_in_8bit": false,
"quant_method": "bitsandbytes"
},
"realtime_patching": true,
"rope_scaling": null,
"rope_theta": 500000.0,
"tie_word_embeddings": false,
"transformers_version": "4.57.3",
"use_cache": false,
"vocab_size": 260
}