|
{ |
|
"_name_or_path": "ibm-granite/granite-20b-code-instruct-8k", |
|
"activation_function": "gelu", |
|
"architectures": [ |
|
"GPTBigCodeForCausalLM" |
|
], |
|
"attention_softmax_in_fp32": true, |
|
"attn_pdrop": 0.1, |
|
"bos_token_id": 0, |
|
"embd_pdrop": 0.1, |
|
"eos_token_id": 0, |
|
"initializer_range": 0.02, |
|
"layer_norm_epsilon": 1e-05, |
|
"model_type": "gpt_bigcode", |
|
"multi_query": true, |
|
"n_embd": 6144, |
|
"n_head": 48, |
|
"n_inner": 24576, |
|
"n_layer": 52, |
|
"n_positions": 8192, |
|
"pad_token_id": 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": true, |
|
"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" |
|
}, |
|
"resid_pdrop": 0.1, |
|
"scale_attention_softmax_in_fp32": true, |
|
"scale_attn_weights": true, |
|
"torch_dtype": "float16", |
|
"transformers_version": "4.47.1", |
|
"use_cache": false, |
|
"vocab_size": 49152 |
|
} |
|
|