ericrisco
/

flor-1-3B-customerservice

Text Generation

customer service

text-generation-inference

Inference Endpoints

4-bit precision

Model card Files Files and versions Community

ericrisco commited on Feb 7

Commit

7ad0c7c

•

1 Parent(s): f7f285e

Update config.json

Files changed (1) hide show

config.json +2 -2

config.json CHANGED Viewed

@@ -22,7 +22,7 @@
   "pad_token_id": 3,
   "pretraining_tp": 1,
   "quantization_config": {
-    "bnb_4bit_compute_dtype": "float32",
     "bnb_4bit_quant_type": "nf4",
     "bnb_4bit_use_double_quant": false,
     "llm_int8_enable_fp32_cpu_offload": false,
@@ -37,7 +37,7 @@
   "skip_bias_add": true,
   "skip_bias_add_qkv": false,
   "slow_but_exact": false,
-  "torch_dtype": "float32",
   "transformers_version": "4.37.2",
   "unk_token_id": 0,
   "use_cache": false,

   "pad_token_id": 3,
   "pretraining_tp": 1,
   "quantization_config": {
+    "bnb_4bit_compute_dtype": "float16",
     "bnb_4bit_quant_type": "nf4",
     "bnb_4bit_use_double_quant": false,
     "llm_int8_enable_fp32_cpu_offload": false,
   "skip_bias_add": true,
   "skip_bias_add_qkv": false,
   "slow_but_exact": false,
+  "torch_dtype": "float16",
   "transformers_version": "4.37.2",
   "unk_token_id": 0,
   "use_cache": false,