r2 / config.json
snoopsy's picture
Run 8. Outer Step 66. Inner Step 135.
f8ec1ad verified
{
"_name_or_path": "snoopsy/r2",
"all_reduce_scores": {
"0": "NOT_ALIVE",
"1": "FAIL",
"10": "NOT_ALIVE",
"100": "NON_PARTICIPATING",
"101": "FAIL",
"102": "NOT_ALIVE",
"103": "NON_PARTICIPATING",
"104": "NON_PARTICIPATING",
"105": "NON_PARTICIPATING",
"106": "SUCCESS",
"107": "NON_PARTICIPATING",
"108": "FAIL",
"109": "FAIL",
"11": "SUCCESS",
"110": "NON_PARTICIPATING",
"111": "NON_PARTICIPATING",
"112": "FAIL",
"113": "NOT_ALIVE",
"114": "NOT_ALIVE",
"115": "FAIL",
"116": "NON_PARTICIPATING",
"117": "NON_PARTICIPATING",
"118": "NON_PARTICIPATING",
"119": "FAIL",
"12": "NOT_ALIVE",
"120": "NON_PARTICIPATING",
"121": "FAIL",
"122": "FAIL",
"123": "FAIL",
"124": "NOT_ALIVE",
"125": "NOT_ALIVE",
"126": "FAIL",
"127": "FAIL",
"128": "SUCCESS",
"129": "NON_PARTICIPATING",
"13": "FAIL",
"130": "NON_PARTICIPATING",
"131": "NON_PARTICIPATING",
"132": "FAIL",
"133": "NON_PARTICIPATING",
"134": "SUCCESS",
"135": "FAIL",
"136": "NON_PARTICIPATING",
"137": "NON_PARTICIPATING",
"138": "NOT_ALIVE",
"139": "FAIL",
"14": "NOT_ALIVE",
"140": "SUCCESS",
"141": "NOT_ALIVE",
"142": "SUCCESS",
"143": "NON_PARTICIPATING",
"144": "NON_PARTICIPATING",
"145": "SUCCESS",
"146": "NON_PARTICIPATING",
"147": "FAIL",
"148": "NON_PARTICIPATING",
"149": "FAIL",
"15": "FAIL",
"150": "FAIL",
"151": "NON_PARTICIPATING",
"152": "NON_PARTICIPATING",
"153": "NON_PARTICIPATING",
"154": "NOT_ALIVE",
"155": "FAIL",
"156": "NOT_ALIVE",
"157": "NON_PARTICIPATING",
"158": "NON_PARTICIPATING",
"159": "NON_PARTICIPATING",
"16": "NOT_ALIVE",
"160": "NON_PARTICIPATING",
"161": "NOT_ALIVE",
"162": "NOT_ALIVE",
"163": "SUCCESS",
"164": "FAIL",
"165": "FAIL",
"166": "NOT_ALIVE",
"167": "FAIL",
"168": "SUCCESS",
"169": "SUCCESS",
"17": "NOT_ALIVE",
"170": "NON_PARTICIPATING",
"171": "SUCCESS",
"172": "FAIL",
"173": "SUCCESS",
"174": "NON_PARTICIPATING",
"175": "FAIL",
"176": "NOT_ALIVE",
"177": "FAIL",
"178": "FAIL",
"179": "FAIL",
"18": "NOT_ALIVE",
"180": "FAIL",
"181": "SUCCESS",
"182": "NOT_ALIVE",
"183": "SUCCESS",
"184": "FAIL",
"185": "SUCCESS",
"186": "SUCCESS",
"187": "NON_PARTICIPATING",
"188": "SUCCESS",
"189": "NON_PARTICIPATING",
"19": "NOT_ALIVE",
"190": "NON_PARTICIPATING",
"191": "FAIL",
"192": "FAIL",
"193": "NON_PARTICIPATING",
"194": "NON_PARTICIPATING",
"195": "SUCCESS",
"196": "FAIL",
"197": "NON_PARTICIPATING",
"198": "NON_PARTICIPATING",
"199": "FAIL",
"2": "NOT_ALIVE",
"20": "NOT_ALIVE",
"200": "NON_PARTICIPATING",
"201": "NOT_ALIVE",
"202": "FAIL",
"203": "FAIL",
"204": "SUCCESS",
"205": "FAIL",
"206": "FAIL",
"207": "NON_PARTICIPATING",
"208": "FAIL",
"209": "NON_PARTICIPATING",
"21": "FAIL",
"210": "NON_PARTICIPATING",
"211": "FAIL",
"212": "NON_PARTICIPATING",
"213": "FAIL",
"214": "FAIL",
"215": "SUCCESS",
"216": "FAIL",
"217": "SUCCESS",
"218": "NOT_ALIVE",
"219": "NON_PARTICIPATING",
"22": "SUCCESS",
"220": "NON_PARTICIPATING",
"221": "FAIL",
"222": "SUCCESS",
"223": "SUCCESS",
"224": "NON_PARTICIPATING",
"225": "FAIL",
"226": "NON_PARTICIPATING",
"227": "NON_PARTICIPATING",
"228": "NON_PARTICIPATING",
"229": "NON_PARTICIPATING",
"23": "NOT_ALIVE",
"230": "FAIL",
"231": "NON_PARTICIPATING",
"232": "NOT_ALIVE",
"233": "FAIL",
"234": "SUCCESS",
"235": "NOT_ALIVE",
"236": "SUCCESS",
"237": "NOT_ALIVE",
"238": "FAIL",
"239": "NON_PARTICIPATING",
"24": "FAIL",
"240": "NON_PARTICIPATING",
"241": "NON_PARTICIPATING",
"242": "FAIL",
"243": "NOT_ALIVE",
"244": "FAIL",
"245": "FAIL",
"246": "NON_PARTICIPATING",
"247": "SUCCESS",
"248": "NON_PARTICIPATING",
"249": "NOT_ALIVE",
"25": "SUCCESS",
"250": "NON_PARTICIPATING",
"251": "NOT_ALIVE",
"252": "FAIL",
"253": "SUCCESS",
"254": "NOT_ALIVE",
"255": "NON_PARTICIPATING",
"26": "FAIL",
"27": "FAIL",
"28": "FAIL",
"29": "SUCCESS",
"3": "SUCCESS",
"30": "FAIL",
"31": "FAIL",
"32": "SUCCESS",
"33": "NOT_ALIVE",
"34": "NON_PARTICIPATING",
"35": "SUCCESS",
"36": "FAIL",
"37": "SUCCESS",
"38": "NON_PARTICIPATING",
"39": "FAIL",
"4": "NOT_ALIVE",
"40": "FAIL",
"41": "FAIL",
"42": "FAIL",
"43": "NON_PARTICIPATING",
"44": "SUCCESS",
"45": "NOT_ALIVE",
"46": "FAIL",
"47": "SUCCESS",
"48": "NOT_ALIVE",
"49": "FAIL",
"5": "FAIL",
"50": "FAIL",
"51": "NON_PARTICIPATING",
"52": "NON_PARTICIPATING",
"53": "FAIL",
"54": "FAIL",
"55": "SUCCESS",
"56": "FAIL",
"57": "NON_PARTICIPATING",
"58": "NOT_ALIVE",
"59": "NOT_ALIVE",
"6": "NOT_ALIVE",
"60": "FAIL",
"61": "NON_PARTICIPATING",
"62": "FAIL",
"63": "SUCCESS",
"64": "SUCCESS",
"65": "NON_PARTICIPATING",
"66": "NON_PARTICIPATING",
"67": "NON_PARTICIPATING",
"68": "SUCCESS",
"69": "NON_PARTICIPATING",
"7": "FAIL",
"70": "FAIL",
"71": "NON_PARTICIPATING",
"72": "FAIL",
"73": "NON_PARTICIPATING",
"74": "SUCCESS",
"75": "FAIL",
"76": "NON_PARTICIPATING",
"77": "NOT_ALIVE",
"78": "NON_PARTICIPATING",
"79": "FAIL",
"8": "NOT_ALIVE",
"80": "NON_PARTICIPATING",
"81": "NON_PARTICIPATING",
"82": "FAIL",
"83": "NOT_ALIVE",
"84": "NON_PARTICIPATING",
"85": "NON_PARTICIPATING",
"86": "NOT_ALIVE",
"87": "FAIL",
"88": "NON_PARTICIPATING",
"89": "NON_PARTICIPATING",
"9": "NOT_ALIVE",
"90": "FAIL",
"91": "NOT_ALIVE",
"92": "FAIL",
"93": "NON_PARTICIPATING",
"94": "NON_PARTICIPATING",
"95": "FAIL",
"96": "NON_PARTICIPATING",
"97": "FAIL",
"98": "NON_PARTICIPATING",
"99": "NON_PARTICIPATING"
},
"architectures": [
"LlamaForCausalLM"
],
"attention_bias": false,
"attention_dropout": 0.0,
"block_list": [
6391291,
6391301
],
"bos_token_id": 1,
"eos_token_id": 2,
"hidden_act": "silu",
"hidden_size": 2048,
"initializer_range": 0.02,
"inner_step": 135,
"intermediate_size": 5632,
"last_allreduce_block": 6386584,
"max_position_embeddings": 2048,
"mlp_bias": false,
"model_type": "llama",
"num_attention_heads": 32,
"num_hidden_layers": 22,
"num_key_value_heads": 4,
"pretraining_tp": 1,
"rms_norm_eps": 1e-05,
"rope_scaling": null,
"rope_theta": 10000.0,
"tie_word_embeddings": false,
"torch_dtype": "float32",
"transformers_version": "4.39.3",
"use_cache": false,
"vocab_size": 32000
}