| { | |
| "_name_or_path": "/tmp/models/tmp3y5bwtdu/tmpj0r4i7dk", | |
| "architectures": [ | |
| "GPTJXForCausalLM" | |
| ], | |
| "auto_map": { | |
| "AutoConfig": "pretrained_config.GPTJXConfig", | |
| "AutoModelForCausalLM": "BeardedMonster/SabiYarn-125M--pretrained_model.GPTJXForCausalLM" | |
| }, | |
| "bias": false, | |
| "block_size": 1024, | |
| "dropout": 0.0, | |
| "model_type": "nanogpt-j", | |
| "n_embd": 768, | |
| "n_head": 12, | |
| "n_layer": 12, | |
| "quantization_config": { | |
| "transformer.h.attn.c_attn": { | |
| "offload_meta": false, | |
| "scale_quant_params": null, | |
| "weight_quant_params": { | |
| "axis": 0, | |
| "channel_wise": true, | |
| "group_size": 64, | |
| "nbits": 4, | |
| "optimize": true, | |
| "round_zero": true, | |
| "view_as_float": false | |
| }, | |
| "zero_quant_params": { | |
| "channel_wise": false, | |
| "group_size": null, | |
| "nbits": 8, | |
| "optimize": false | |
| } | |
| }, | |
| "transformer.h.attn.c_proj": { | |
| "offload_meta": false, | |
| "scale_quant_params": null, | |
| "weight_quant_params": { | |
| "axis": 0, | |
| "channel_wise": true, | |
| "group_size": 64, | |
| "nbits": 4, | |
| "optimize": true, | |
| "round_zero": true, | |
| "view_as_float": false | |
| }, | |
| "zero_quant_params": { | |
| "channel_wise": false, | |
| "group_size": null, | |
| "nbits": 8, | |
| "optimize": false | |
| } | |
| }, | |
| "transformer.h.mlp.c_fc": { | |
| "offload_meta": false, | |
| "scale_quant_params": null, | |
| "weight_quant_params": { | |
| "axis": 0, | |
| "channel_wise": true, | |
| "group_size": 64, | |
| "nbits": 4, | |
| "optimize": true, | |
| "round_zero": true, | |
| "view_as_float": false | |
| }, | |
| "zero_quant_params": { | |
| "channel_wise": false, | |
| "group_size": null, | |
| "nbits": 8, | |
| "optimize": false | |
| } | |
| }, | |
| "transformer.h.mlp.c_proj": { | |
| "offload_meta": false, | |
| "scale_quant_params": null, | |
| "weight_quant_params": { | |
| "axis": 0, | |
| "channel_wise": true, | |
| "group_size": 64, | |
| "nbits": 4, | |
| "optimize": true, | |
| "round_zero": true, | |
| "view_as_float": false | |
| }, | |
| "zero_quant_params": { | |
| "channel_wise": false, | |
| "group_size": null, | |
| "nbits": 8, | |
| "optimize": false | |
| } | |
| } | |
| }, | |
| "torch_dtype": "float32", | |
| "transformers_version": "4.44.0", | |
| "vocab_size": 52050 | |
| } | |