comoZ commited on
Commit
d00f3dd
·
verified ·
1 Parent(s): 03e050c

Upload LlamaForCausalLM

Browse files
Files changed (2) hide show
  1. README.md +0 -2
  2. config.json +1 -1
README.md CHANGED
@@ -140,8 +140,6 @@ extra_gated_heading: Please be sure to provide your full legal name, date of bir
140
 
141
  **Tri-21B**를 4bit으로 양자화한 모델
142
 
143
- https://huggingface.co/trillionlabs/Tri-21B
144
-
145
  We introduce **Tri-21B**, our flagship large language model that redefines the efficiency frontier in LLM training. By achieving state-of-the-art performance with only 2.3T training tokens, we demonstrate that exceptional capabilities don't require excessive computational resources.
146
 
147
  <p align="center">
 
140
 
141
  **Tri-21B**를 4bit으로 양자화한 모델
142
 
 
 
143
  We introduce **Tri-21B**, our flagship large language model that redefines the efficiency frontier in LLM training. By achieving state-of-the-art performance with only 2.3T training tokens, we demonstrate that exceptional capabilities don't require excessive computational resources.
144
 
145
  <p align="center">
config.json CHANGED
@@ -23,7 +23,7 @@
23
  "quantization_config": {
24
  "_load_in_4bit": true,
25
  "_load_in_8bit": false,
26
- "bnb_4bit_compute_dtype": "float16",
27
  "bnb_4bit_quant_storage": "uint8",
28
  "bnb_4bit_quant_type": "nf4",
29
  "bnb_4bit_use_double_quant": false,
 
23
  "quantization_config": {
24
  "_load_in_4bit": true,
25
  "_load_in_8bit": false,
26
+ "bnb_4bit_compute_dtype": "float32",
27
  "bnb_4bit_quant_storage": "uint8",
28
  "bnb_4bit_quant_type": "nf4",
29
  "bnb_4bit_use_double_quant": false,